PMRes 2023 4

Institute
of
Naval architecture
No 4 (120) 2023
Vol. 30
ADDRESS OF PUBLISHER
& EDITOR’S OFFICE:
GDAŃSK UNIVERSITY
OF TECHNOLOGY
Institute
of Naval Architecture 4 Piotr Gomułka
G. Narutowicza 11/12 Computer-aided system for layout of fire hydrants on boards
80-233 Gdańsk, POLAND designed vessel using the Particle Swarm Optimization algorithm
17 Yordan Denev
RETROFITTING THE BOW OF A GENERAL CARGO VESSEL ANDEVALUATING
ENERGY EFFICIENCY OPERATIONAL INDEX
24 Jan Sierzputowski, Artur Karczewski, Przemysław Krata
USE OF THE AHP METHOD FOR PREFERENCE DETERMINATION IN YACHT
EDITORIAL STAFF: DESIGN
31 Nguyen Dinh Thach, Phan Van Hung
Wiesław Tarełko OPTIMAL UV QUANTITY FOR A BALLAST WATER TREATMENT SYSTEM FOR
| Editor in Chief COMPLIANCE WITH IMO STANDARDS
Janusz Kozak 43 Se-yeol An, Hyeon-won Jeong, Ohyoung Kim, W. Jaewoo Shim
| Deputy Editors-in-Chief EFFECTS OF SWAY AND ROLL EXCITATIONS ON SLOSHING LOADS IN A KC-1
Wojciech Litwin MEMBRANE LNG TANK
| Deputy Editors-in-Chief 54 Jacek Nakielski
ANALYSIS OF THE ENVIRONMENTAL IMPACT OF THE HULL CONSTRUCTION
OF A SMALL VESSEL BASED ON LCA
61 Ali Zinati, Mohammad Javad Ketabdari, Hamid Zeraatgar
EFFECTS OF PROPELLER FOULING ON THE HYDRODYNAMIC PERFORMANCE
Price: OF A MARINE PROPELLER
single issue: 25 PLN 74 Volodimir Leshchev, Igor Maslov, Oleksandr Palagin, Andrii Naydyonov
TRANSFER FUNCTION FOR A CONTROLLABLE PITCH PROPELLER WITH ADDED
Prices for abroad WATER MASS
single issue: 81 Dmytro Minchev, Roman Varbanets, Oleksandr Shumylo, Vitalii Zalozh,
- in Europe EURO 15 Nadiia Aleksandrovska, Pavlo Bratchenko, Thanh Hai Truong
- overseas USD 20 Digital Twin test-bench Performance for marine diesel engine
applications
WEB: 92 Gang Wu, Guodong Jiang, Changsheng Chen, Guohe Jiang, Xigang Pu,
pg.edu.pl/pmr Biwen Chen
AN EXPERIMENTAL STUDY OF THE EFFECTS OF CYLINDER LUBRICATING
e-mail : pmr@pg.edu.pl OILS ON THE VIBRATION CHARACTERISTICS OF A TWO-STROKE LOW-SPEED
MARINE DIESEL ENGINE
ISSN 1233-2585
ADDRESS OF PUBLISHER 102 Volodymyr Korobko, Serhiy Serbin, Huu Cuong Le
& EDITOR’S OFFICE: EXPLORATION OF A MODEL THERMOACOUSTIC TURBOGENERATOR WITH
A BIDIRECTIONAL TURBINE
GDAŃSK UNIVERSITY 110 Jerzy Girtler, Jacek Rudnicki
OF TECHNOLOGY QUANTUMNESS IN DIAGNOSTICS OF MARINE INTERNAL COMBUSTION
ENGINES AND OTHER SHIP POWER PLANT MACHINES
Institute 120 Piotr Bilon, Wojciech Litwin
of Naval Architecture HUMAN RESOURCE MANAGEMENT DIGITALISATION IN MULTIDISCIPLINARY
G. Narutowicza 11/12 SHIP DESIGN COMPANIES
80-233 Gdańsk, POLAND 129 Minh Duc Nguyen, Ko Tae Yeon, Krzysztof Rudzki, Hoang Phuong Nguyen,
Nguyen Dang Khoa Pham
STRATEGIES FOR DEVELOPING LOGISTICS CENTRES: TECHNOLOGICAL
TRENDS AND POLICY IMPLICATIONS
Editorial
POLISH MARITIME RESEARCH is the scientific journal with a worldwide circulation. This journal is published
quarterly (four times a year) by Gdansk University of Technology (GUT). On September, 1994, the first issue of
POLISH MARITIME RESEARCH was published. The main objective of this journal is to present original research,
innovative scientific ideas, and significant findings and application in the field of :
Naval Architecture, Ocean Engineering and Underwater Technology,
The scope of the journal covers selected issues related to all phases of product lifecycle and corresponding technologies
for offshore floating and fixed structures and their components.
All researchers are invited to submit their original papers for peer review and publications related to methods of the
design; production and manufacturing; maintenance and operational processes of such technical items as:
• all types of vessels and their equipment,
• fixed and floating offshore units and their components,
• autonomous underwater vehicle (AUV) and remotely operated vehicle (ROV).
We welcome submissions from these fields in the following technical topics:
• ship hydrodynamics: buoyancy and stability; ship resistance and propulsion, etc.,
• structural integrity of ship and offshore unit structures: materials; welding; fatigue and fracture, etc.,
• marine equipment: ship and offshore unit power plants: overboarding equipment; etc.
Scientific Board
Chairman : Prof. JERZY GIRTLER - Gdańsk University of Technology, Poland
Vice-chairman : Prof. CARLOS GUEDES SOARES, Universidade de Lisboa, Lisbon, Portugal
Vice-chairman : † Prof. MIROSŁAW L. WYSZYŃSKI - University of Birmingham, United Kingdom
Prof. POUL ANDERSEN Prof. YORDAN GARBATOV Prof. JERZY MERKISZ

Technical University of Denmark Universidade de Lisboa, Poznan University of Technology
Kongens Lyngby Lisbon Poznan
Denmark Portugal Poland
Prof. JIAHN-HORNG CHEN Prof. STANISLAW GUCMA Prof. VALERI NIEKRASOV

National Taiwan Ocean University Maritime University of Szczecin Admiral Makarov National University
Keelung Szczecin of Shipbuilding
Taiwan Poland Mikolaiv
Ukraine
Prof. VINCENZO CRUPI Prof. JIAHN-HORNG CHEN
University of Messina National Taiwan Ocean University Prof. SERHIY SERBIN
Messina Keelung Admiral Makarov National
Italy Taiwan University of Shipbuilding
Mikolaiv
Prof. MAREK DZIDA Prof. ANTONI ISKRA Ukraine
Gdansk University of Technology Poznan University of Technology
Gdansk Poznan Prof. JOZEF SZALA
Poland Poland UTP University of Science and
Technology
Dr. KATRIEN ELOOT, Prof. JAN KICINSKI Bydgoszcz
Flanders Hydraulics Research, Institute of Fluid-Flow Machinery - Poland
Antwerpen Polish Academy of Sciences
Belgium Gdansk Prof. HOANG ANH TUAN
Poland Ho Chi Minh City
Prof. ODD MAGNUS FALTINSEN University of Technology
Norwegian University of Science and Prof. ZBIGNIEW KORCZEWSKI (HUTECH)
Technology Gdansk University of Technology Ho Chi Minh
Trondheim Gdansk Vietnam
Norway Poland
Prof. TADEUSZ SZELANGIEWICZ
Prof. MASSIMO FIGARI Prof. JOZEF LISOWSKI Maritime University of Szczecin
University of Genova Gdynia Maritime University Szczecin
Genova Gdynia Poland
Italy Poland
Prof. DRACOS VASSALOS
Prof. HASSAN GHASSEMI Prof. JERZY EDWARD MATUSIAK University of Strathclyde
Amirkabir University of Technology Aalto University Glasgow
Tehran Espoo United Kingdom
Iran Finland
POLISH MARITIME RESEARCH 1 (117) 2023 Vol. 30; pp. 4-16
10.2478/pomr-2023-0053
Computer-aided system for layout of fire hydrants

on boards designed vessel using the Particle Swarm
Optimization algorithm
Piotr Gomułka
1
Damen Engineering Gdansk, Gdansk
2
Gdansk University of Technology, Gdansk, Poland
* Corresponding author: piotr.gomulka@pg.edu.pl (Piotr Gomułka)
Abstract
The functional layout of fire safety equipment in technical spaces of ships is a time-consuming process. When designing
a ship fire protection system, the designer must manually position each system component in such a way as to meet the
requirements of regulations arising from the technical specification, various legal regulations of maritime conventions
and classification societies of the vessel to be designed. Layout of fire hydrants assisted by a computer that is based
on pre-defined criteria and various constraints could significantly support the designer in working easier and faster.
This paper presents a prototype computer-aided design system that enables optimal placement of fire hydrants using
the metaheuristic Particle Swarm Optimization (PSO) algorithm. This algorithm was used in Rhinoceros 3D software
with its Grasshopper plugin for visualizing the arrangement of fire safety equipment. Various solution arrangements
compared with the fire hydrant placement in real ships are illustrated by a case study. Demonstrating how design work
can be facilitated and what potential benefits can be achieved are presented as well.
Keywords: ship, fire hydrant, design, layout, particle swarm optimization
introduction FFS Resolution2 , classification society rules, and standards

that specify requirements for the designed fire protection
The placement of fire protection equipment in ship installation, as well as the technical specification of the vessel.
technical rooms is an extremely important and time- Then, based on legal and technical requirements and hazard
consuming part of the ship design. One of the important analysis, the designer begins to layout hydrants on a ship’s
components of this equipment is the fire hydrant, and board, including determining their minimum number
its rational or optimal placement on board ships should and location to ensure maximum effectiveness of the fire
reduce the degree of fire hazard. The proper placement of protection system. This takes into account factors such as the
fire hydrants requires designers to conduct a preliminary type of hydrants, water flow and pressure, and ease of access
analysis of the fire hazards that may occur on the ship’s to hydrants. In existing ship design practice, the designer
board. This should be preceded by familiarization with the manually positions each fire hydrant to meet the imposed
applicable regulations, such as the SOLAS Convention1, the requirements based on his experience and knowledge, use
a database of similar design solutions, and also follow the
guidelines of installed equipment manufacturers. As a rule,
1
The International Convention for the Safety of Life at Sea (SOLAS)–
an international maritime treaty that sets minimum safety standards 2
The Fire Safety Systems (FSS) Code – an international standard
for ships, including requirements for design, construction, equipment, that provides guidelines for the design, installation, and maintenance
and operation. of fire safety systems on board ships.
4 POLISH MARITIME RESEARCH, No 4/2023

the manufacturers specify, for example, the necessary service level α. In addition, the authors model the maximal covering
or operator space, the location of connecting cooperating location problem in an uncertain environment using different
installations, or impose the order of their installation. The last modeling ideas, namely, the (α, β)-maximal covering location
step is calculations or simulations regarding the hydraulics model and the α-chance maximal covering location model.
of the system performed to ensure that each hydrant will It is also proven that the (α, β)-maximal covering location
have adequate water flow and pressure and that the entire model can be transformed into an equivalent deterministic
fire protection system will work effectively in the event of location model, and then, it can be solved. They also point
a fire occurrence. out that there exists an equivalence relation between the
This paper presents a prototype computer-aided design (α, β)-maximal covering location model and the α-chance
system that enables optimal placement of fire hydrants maximal covering location model, which leads to a method
using the metaheuristic Particle Swarm Optimization (PSO) for solving the α-chance maximal covering location model.
algorithm. The PSO algorithm is embedded in Rhinoceros Finally, the ideas of uncertain models are illustrated by a case
3D software with its Grasshopper plugin for visualizing the study.
arrangement of fire protection equipment. The Grasshopper There are also a number of studies presenting solutions
program is a visual scripting environment for Rhinoceros 3D for facilitating the placement of equipment or installations in
software that is very popular in ship design offices. Users can ship technical rooms. For example, criteria for evaluating the
both build and use a library of existing parametric algorithms placement of equipment in the ship engine room and analysis
to modify, analyze, or create 3D models from scratch. They methods that allow for choosing the best option in terms
are also able to automate design processes without the need of safety, energy efficiency, and costs are presented in [6].
for writing software code. On the other hand, methods and technologies that improve
Various solution arrangements compared with the fire access to equipment in the engine machinery room in terms
hydrant placement in real ships are illustrated by a case study. of maintenance cost planning and effective maintenance
Demonstrating how design work can be facilitated and what strategy implementation are presented in [7]. The study [8]
potential benefits can be achieved are presented as well. concerns the integration of computer-aided and knowledge
base systems in designing ship machinery equipment and
installations. The authors describe how integrating these
Literature review two systems allows for better optimization and increases the
efficiency of the design process that ensures greater device
In the subject literature, there are many concepts and reliability or facilitating their operation. In [9], the authors
solutions for object placement in any enclosed space. For focus on issues related to the design of ship pipeline systems.
example, in [1], the authors presented three approaches to For example, they discuss pipeline layout design criteria,
automating the object placement process: rule-based, genetic taking into account aspects related to the difficult working
algorithms, and artificial neural networks. They also discuss conditions, such as vibration or corrosion.
ways to improve the quality of automatic object placement Despite the numerous articles on this topic, simple
by taking into account the specificities of space and user engineering decision support systems enabling the generation
preferences. The authors of the article referenced in [2] of alternative layouts of equipment, including the placement
described a new approach to automatic object positioning of fire safety equipment on board ships, have not been
in three-dimensional spaces. The method described there presented. This is confirmed, among others, by authors of
is based on the use of constraints that allow for precise [10], stating that this issue is not well researched in terms of
control of the object placement process in space. They general solutions. They propose to solve that problem using
present a mathematical model that allows for the formal the a modified iterative-deepening search method. It is the
definition of constraints and their application in the object iterative-exploration method that uses a classical greedy
positioning process. In the publication [3], the authors provide algorithm by means of it is possible to show the possibility
an overview of problems related to object location that have of preliminary placement of fire safety equipment.
applications in various fields, including transport network Our approach is based on developing a prototype
planning, store location, warehouse, and medical centers. computer-aided design system that enables optimal placement
On the other hand, the possibilities of using tools based of fire hydrants using the metaheuristic Particle Swarm
on various variants of p-median and p-center algorithms Optimization (PSO) algorithm.
to solve transport network planning problems, store and
distribution center location are described in [4]. The paper Concept of computer-aided design system
[5] employs uncertainty theory to address the location that enables optimal placement of fire
problem of emergency service facilities under uncertainty. hydrants
Using the inverse uncertainty distribution, the uncertain
location set covering model was transformed into an PSO algorithm as a means of supporting the placement of
equivalent deterministic location model. This paper first fire hydrants on ship decks
studies the uncertainty distribution of the covered demand To solve the problem of optimal layout of fire hydrants
that is associated with the covering constraint confidence on the shipboard, one type of meta-heuristic algorithm, i.e.,
POLISH MARITIME RESEARCH, No 4/2023 5

the PSO algorithm, was used in the developed computer- methods that measure the efficiency of algorithms (e.g.,
aided system. It has been found to be particularly effective in convergence time, stability, resistance to getting stuck in
optimizing problems related to closed spaces because it can local optima).
efficiently explore and exploit the search space. Closed spaces, In this study, the goal is not to compare individual
such as those found structural optimization problems, often algorithms in terms of their speed or effectiveness but to
have complex constraints and interactions between variables demonstrate that the use of simple optimization tools in the
that can make classical optimization methods difficult to placement of fire protection system components can bring
apply [11]. However, the PSO algorithm is able to navigate noticeable benefits to the designer (reducing design time),
these complex search spaces by simultaneously exploring the shipyards (lowering installation costs), and ultimately ship-
search space and exploring promising regions of the search owners, who will be responsible for maintaining the selected
space. installation components. Based on the criteria of ease of
Additionally, the PSO algorithm is capable of handling implementation with the system responsible for placing the
non-linear and non-convex optimization problems, which are fire protection system components, as well as the simplicity
often encountered in closed-space optimization problems. By of its operation and use, the PSO algorithm was used, which
using a swarm of particles, the PSO can avoid getting stuck works by using the “Silvereye” plugin.
in local optima and instead converge to a global optimum At its core, the PSO involves a population of candidate
solution. However, the effectiveness of the algorithm depends solutions, called particles, moving around in a search space
on various factors such as the complexity of the problem, the and adjusting their position based on their own best-known
size of the space, and the number of design variables involved. position and the best-known position of their neighbors.
Therefore, it is important to carefully evaluate the suitability Mathematically, the position of each particle in the search
of the PSO algorithm for a specific application related to space is represented by a vector x, and its velocity is represented
closed spaces before implementation. by a vector v. The objective function that needs to be optimized
However, it should be noted that PSO, as a member of is denoted as f(x). The PSO algorithm iteratively updates the
the metaheuristic methods family, only provides a way to position and velocity of particles until a termination criterion
create an appropriate heuristic algorithm. In turn, such an is met, such as a maximum number of iterations or a target
algorithm enables the obtainment of a solution for which it objective function value. The mathematical model of the
is possible to prove how close it is to the optimal solution, PSO algorithm has been found in many studies, e.g., in [19],
i.e., it is a quasi-optimal solution. From the point of view of [20], and [21].
engineering practice and ship design offices, such a solution To develop a prototype computer-aided design system that
is fully acceptable. enables optimal placement of fire hydrants on ship-boards,
As previously mentioned, the PSO algorithm is built into we should formulate an optimization problem. To formulate
the Grasshopper application and is an integral component it using the PSO algorithm in Grasshopper, we can follow
of it. The useful features of the Grasshopper application these two steps:
were used to create a tool that assists designers in properly –– define the problem by defining the constraints, the
placing fire hydrants. By defining sets of design rules and objective functions, and the design variables that can be
constraints that determine the placement of these hydrants, adjusted to optimize the objective functions,
an algorithm was chosen that is suitable for the complexity –– use a PSO algorithm implementation in Grasshopper using
of the computational problem. the “Silvereye” plugin.
Meta-heuristic algorithms, while not guaranteeing the
discovery of global optimal solutions, can provide results close Constraints
to optimal in a reasonable amount of time. This seems to be To ensure optimal placement of hydrants on board
a favorable solution for placing fire hydrants on a shipboard. a designed ship, a designer support system should take into
Meta-heuristics are often inspired by natural processes, such as account the following constraints:
swarm interactions (particle swarm optimization, ant colony –– space constraints; the available space on the ship deck is
optimization), generational evolution (genetic algorithms, limited by the size and shape of the ship,
genetic programming, evolutionary programming), as well as –– legal regulations; they dictate the minimum number and
physical phenomena (e.g., simulated annealing). Among the placement of hydrants on ship decks,
existing optimization tools in the Grasshopper environment, –– significance of compartments; areas on the ship where fire
frequently used ones include Galapagos [12], [13], Goat [14], is most likely or the risk is negligible.
Silvereye [15], Opossum [16], Dodo [17], and Nelder Mead
[18]. However, choosing the best one is a very difficult and Space constraints
ambiguous task. In the subject literature, both simulation The developed system that enables optimal placement of
studies have shown the superiority of individual algorithms hydrants has to take into account the physical limitation
[14], [15] and works that suggested that no single algorithm of the vessel’s hull and the shape of the defined spaces. The
was dominant for the considered optimization problems shape of the hull can take on various forms, depending on
[14]. Therefore, a clear assessment of the usefulness of the the level of detail and complexity of the model. In Rhino Core
tool depends on the optimization problem and comparative software for design and modelling, it can be represented by:

–– curves of various types, such as NURBS (Non-Uniform where:
Rational B-Splines), Bezier, interpolating, spline, pi is the i-th control point,
modeling,– – surface models, such as NURBS, which are one of the𝑤𝑤𝑖𝑖 is theWweight
basic i
is thecoefficient
weight coefficient of the
of the 𝑖𝑖-th i-th control
control point,
point, determining the in
types of surfaces supported by Rhino Core, allowing for determining the influence of each of the control points on
importedprecise
from definition
external ofCAD the shape of the hullincluding
file formats, using control
of the
IGES the
control curve,
(Initialpoints on the curve,
Graphics
points and curves, Bin is the i-th degree Bernstein polynomial expressed by the
–– mesh surface
ge Specification), STEP (Standard for the Exchange𝐵𝐵𝑖𝑖𝑛𝑛ofisProduct
models, 𝑛𝑛-thmodel
theformula: degreedata),
Bernstein polynomial expressed by the formula:
–– hybrid modeling,
tandard––ACIS
models Text), and others;
imported the hullCAD
from external can file
be designed
formats, in another program
𝑛𝑛 (𝑛𝑛𝑖𝑖)𝑡𝑡 𝑖𝑖 (1 − 𝑡𝑡)𝑛𝑛−𝑖𝑖 𝑓𝑓𝑓𝑓𝑓𝑓 𝑖𝑖 = 0, … , 𝑛𝑛
including IGES (Initial Graphics Exchange Specification),
ported into Rhino Core, where further work on the project can be𝑖𝑖 carried out. 𝐵𝐵 (𝑡𝑡) = { (2)
STEP (Standard for the Exchange of Product model data), 0 𝑓𝑓𝑓𝑓𝑓𝑓 𝑖𝑖 < 0, 𝑖𝑖 > 𝑛𝑛
ed system,SAT each(Standard ACISrooms
of the ship Text), and
wasothers;
created theusing
hull can be curves (polylines),
closed
designed in another program and imported into Rhino Legal regulations
he outline Core,
of awhere further space.
particular work on The
the project can beassembled
outlines carried out. together According
form ato flat
the SOLAS convention, the number and
In the developed system, each of the ship rooms was regulations
Legal created placement of hydrants should be such that at least two water
ach of the
usinganalyzed vessel's
closed curves decks, representing
(polylines), as shown inthe Figure
outline 1.ofSuchstreams,
outlines notcan also from the same hydrant, one of which
originating
According to theshould
SOLAS convention, the number and placement of hydrants
xample,aon particular space. The
the general planoutlines
of theassembled together
vessel made in aform
CAD a flat
program. be supplied by a single length of hose, can reach any
projection on each of the analyzed vessel’s decks, as shown part of the ship accessible to passengers or crew during ship
that at least two navigation.
in Figure 1. Such outlines can also be based, for example,
water streams, not originating from the same hydrant, one of w
Fire hoses should be long enough to cover any
on the general plan of the vessel made in a CAD program.
supplied compartment
by a single length ofbyhose, a water
canstream
reachwhere theirof
any part use
themay be accessible t
ship
crew during ship navigation. Fire hoses should be long enough to cover any c
a water stream where their use may be required. Fire hoses should be at least 1
but no more than: 15 meters in engine rooms, 20 meters in other compartmen
decks, and 25 meters on open decks of ships with a maximum width exceeding
In the developed system, the range of the fire hose creates a circle with a radi
length of the hose defined in the SOLAS convention. It should be noted that
Example top view projection of ship rooms created using closed curves
simplification, as the actual range of the fire hose may be slightly smaller
complicated
Fig. 1. Example top view projection of ship rooms created using closed curves compartment configuration and the need for its bends. In the dev
hould be located within the ship's geometrical representation, required. Fire hoses should be at least 10 meters long, but
and any obstacles
the hose length can no more than: 15
be easily meters and
defined in engine rooms,
adjusted to20
anymeters in otherconstraints o
additional
Firereinforcements)
e.g., pillars, hydrants shouldmust be located
be taken within the ship’sand compartments
into account and on open decks, and 25 meters on open
should not create
geometrical representation, and any obstacles e.g., imposed
on the stateofauthority
vessel bydecks ships with aregulations,
maximum width which is illustrated
exceeding in Figure 2.
30 meters.
In rooms, hydrants
(e.g., are usually located
pillars, reinforcements) must near walls,
be taken intotypically
account near communication
In the developed system, the range of the fire hose creates
and should
ation exits. In ournot create anyitcollisions.
approach, was assumedIn rooms,
thathydrants are
fire hydrantsawould
circle with
be alocated
radius equal to the length of the hose defined in
usually located near walls, typically near communication the SOLAS convention. It should be noted that this is a certain
routes the
rve, forming or evacuation exits.
outline of theInselected
our approach,
room.it Any
was assumed simplification,
point markers as the actual range of the fire hose may be
on the curve
that fire hydrants would be located on a closed curve, forming slightly smaller in the case of complicated compartment
he graphic representation
the outline of a room.
of the selected hydrant.
AnyB-spline curves
point markers are used
on the to interpolate
configuration a need for its bends. In the developed
and the
curve can represent the graphic representation of a hydrant. system, the hose length can be easily defined and adjusted to
d by 𝑁𝑁 knot points. They are a chain of Bezier curves. Any pointany
B-spline curves are used to interpolate a curve defined by
onadditional
a polynomial
constraints or requirements, e.g., imposed by
knot
oordinates ispoints. They areby
determined a chain of Bezier curves.
the following Any point on
equation: state authority regulations, which is illustrated in Figure 2.
a polynomial curve in real coordinates is determined by the
following equation:
𝑛𝑛
∑𝑖𝑖=0 𝑤𝑤𝑖𝑖 ∙𝑝𝑝𝑖𝑖 ∙𝐵𝐵𝑖𝑖𝑛𝑛 (𝑡𝑡)
𝑝𝑝(𝑡𝑡) = 𝑛𝑛 𝑓𝑓𝑓𝑓𝑓𝑓 𝑡𝑡 𝜖𝜖 (0,1) (1) (1)
∑𝑖𝑖=0 𝑤𝑤𝑖𝑖 ∙𝐵𝐵𝑖𝑖𝑛𝑛 (𝑡𝑡)
e 𝑖𝑖-th control point,

Fig. 2 Possible parameterization options for the length of a fire hose according
convention
Fig. 2. Possible parameterization options for the length of a fire hose according to the SOLAS convention
Significance of compartments electrical equipment, etc. – fire hoses with a length of not
The SOLAS Convention has distinguished different less than 10 [m] and not more than 15 [m] must be used,
classes of bulkheads based on their ability to withstand –– areas on the ship where the probability of fire is slight
fire and smoke on the endangered side. These bulkheads are usually considered places where the risk of fire is
are classified as A, B, and C classes and create thermal and small and negligible, usually with limited human such
structural boundaries on the vessel. Spaces are separated by as voids, cofferdams, tanks, chain lockers, fixed gas fire-
the appropriate class of bulkhead (either bulkheads or decks) extinguishing system storage rooms, and others – in this
depending on the level of fire risk. Details specifying the fire given case, there is no need to install a fire hydrant,
sealing standards to be applied to specific bulkheads between –– other spaces and open decks – fire hoses with a length of
neighboring compartments can be found in SOLAS II-2 Part not less than 10 [m] and not more than 20 [m] must be used.
C. The following types of spaces are most relevant in terms In the developed system, each type of space is assigned to
of fire hydrant distribution: a specific drawing layer (Fig. 3). The number of layers and
–– machinery space category A, which is classically considered their assigned properties can be changed as desired. For
the area with the highest fire risk and requires the highest example, machinery spaces of category A can be assigned
safety standards – fire hoses with a length of not less than to layer 1, while other machinery spaces can be assigned to
10 [m] and not more than 15 [m] must be used, layer 2. Spaces with low fire risk, such as voids or cofferdams,
–– machinery spaces, which include both the previously can be assigned to layer 3. Other spaces and open decks,
mentioned machinery space category A and other spaces where fire hoses are required, can be assigned to layer 4. By
containing propulsion machinery, boilers, fuel units, steam using different layers, the design team can easily distinguish
and internal combustion engines, generators, and main between different types of rooms and apply different design
criteria or safety measures accordingly.
Fig. 3. Possible parameterization options for the length of a fire hose according to the SOLAS convention

Objective functions distribution of hydrants are stored. Then, another simulation
In the case of optimizing the placement of fire hydrants, is carried out reducing the number of hydrants in the next
the objective function determines how well our system works step (manually) until the n-tℎ simulation finds a satisfactory
from the perspective of the optimization criterion(s) chosen solution.
by the designers. They can prioritize each of these criteria, In the case of the first criterion, which is the maximization
and then, the objective function will determine how well we of the coverage area of fire extinguishing, the area of such
meet these requirements. In the case of a few criteria selected, a zone depends on: the number of hydrants, the location of
the objective function can be defined as a weighted sum of hydrants, the length of fire hoses connected to hydrants,
each criterion, where the weights are set by the designer. and the configuration (shape) of the extinguished room. The
In ship design practice, a number of alternative criteria for measure of effectiveness in this case is to have a distribution
the rational placement of fire hydrants on shipboards can be of hydrants that ensures the most efficient fire protection. This
found. Here are the most commonly encountered: can be expressed by the degree of coverage of the extinguished
–– maximization of the coverage of extinguished surfaces, area, also defined as the area within the range of the fire
–– minimization of power consumption to increase fire safety hose. The larger the uncovered area, where there is no range
system efficiency, of action of the fire hoses, the lower the effectiveness of the
–– minimization of risk by optimal placement of hydrants in fire protection system.
areas most vulnerable to fire,
–– maximization of accessibility to hydrants,
–– minimization of response time by optimal placement of
hydrants that allow for fast and effective intervention,
–– minimization of the number of hydrants to minimize costs
and maintain system simplicity,
–– minimization of response time, which assumes minimizing
the time needed to reach the most fire-prone areas on the
ship,
–– minimization of the distance between hydrants to
minimize costs.
It is obvious that the choice of criterion/criteria depends
on many factors. The most important of these include the
type of ship (cargo, passenger, etc.) and the size associated
with it, as well as the preferences of the ship’s owner or future
ship operator. Fig. 4. Layout of hydrants with a demonstration of hose coverage in a selected
In the developed system, the objective functions can be room, taking into account the location of the fire pump, doors, and the area
not covered by the installation
defined by four criteria, which are most preferred by the ship
contractor, namely: One possible way to present the distribution of hydrants
–– the criterion of maximum coverage of the extinguished and the coverage area of hoses in a selected room is to use
surface (or alternatively: minimizing the areas without a floor plan or a map of the room (Fig. 4). The hydrants can
reach of fire hoses), be marked with symbols or icons, and the coverage area
–– the criterion of minimal overlap of water streams, of hoses can be presented as circles around each hydrant,
–– the criterion of minimum power demand for fire pump showing the maximum distance the hoses can reach from
engines, each hydrant. The pump room and other important features
–– the criterion of minimum distance from evacuation exits. such as doors and windows can also be marked on the floor
In all the cases of the objective function considered next, plan. To illustrate the effectiveness of the fire protection
the design variables are the coordinates of the location of the system, the areas that are not covered by the hoses can be
hydrants x and y, which are distributed in the two-dimensional highlighted, indicating the potential areas where the fire
space determined by space conistraints, considering their could spread if the system is not sufficient. This can help to
position close to a closed curve forming the outline of the identify any potential weaknesses in the fire protection system
selected room. These coordinates are the centers of circles, and inform the decision-making process in optimizing the
whose radii are the lengths of fire hoses determined by the placement of hydrants and fire hoses for maximum coverage.
restrictions presented in the legal regulations and significance The developed system utilizes an algorithm that employs
of compartments, respectively. the region union and region difference functions to compare
In addition, the designer arbitrarily determines the number specified surfaces, calculate their difference, and determine
of hydrants placed on each deck or room of the ship. At the their total area. The region union function is used to
current stage of system development, the determination of the combine geometric regions represented as sets of points in
number of hydrants is done iteratively, i.e., the designer takes two-dimensional space that can take different shapes, e.g.,
any number of hydrants and then runs simulations looking polygons. This function performs the operation of sum of
for the best match. If a solution is found – the number and sets, which can be mathematically described as:

Figure 5 shows the results of a simulation in which the positions of three hydrants w
The algorithm placed the hydrants in such a way that the areas not covered by the h
ltant area 𝐴𝐴 R₁
is equal
R₂ ...
to the
R nsum x
= {x :of or x were
 R₂...as
R₁ individual
the or x 𝑅𝑅₁,
small
sets Rn}.𝑅𝑅₂, . . . , 𝑅𝑅ₙ and
asThis
possible. In𝐴𝐴the
Figure 5=shows
𝑅𝑅₁ ∪
case of𝑅𝑅₂
the the∪second
results criterion,inwhich
of a simulation is the min
which the
means that the sum of the sets R₁  R ...  Rn contains those positions of three hydrants were changed. The algorithm
∪ 𝑅𝑅ₙ (𝑛𝑛 is the number
elements ofbelong
x that areas). Any
to the setpoint
overlap
R (x that
of water
R₁) orbelongs
1
belong toto
streams,one
theatsetleastplaced
theofalgorithm
thethese
works
areas
hydrants willon
in such
the same principle as the one ad
a way that the areas not covered
R2 (x  R2) ... or belong to the set Rn (xfirst
 Rn).criterion.
The set of points by the hydrant
The difference, range were
however, is as small
that theas areas
possible. In the caseby
protected of the fi
belong to the resultant area 𝐴𝐴. When the region union function
of the resultant area A is equal to the sum of the individual
runs, the program checks
the second criterion, which is the minimization of overlap of
sets R₁, R₂,...,R
ch areas intersect and thenn
A = R₁  R₂ ...
and combines them  Rto
n
(nform
compared, is the number
taking
a unifiedof area.
into waterIfstreams,
account their
the the algorithm
common
regions have works
parts, on overlapping
i.e., the same principle as 𝐵𝐵a . In t
areas
areas). Any point that belongs to at least one of these areas will the one adopted in the first criterion. The difference, however,
also belong
mon boundaries, thesetoboundaries
the resultantwill the aim
areabeA.included
When isregion
thein to minimize
the the
union union
resulting degree
is that of The
the areas
region. mutual coverage
protected
region of hoses
by the fire the hose ranges of their o
are compared,
function runs, the program checks which areas intersect and taking into account their common parts, i.e., overlapping
erence function
thenis used tothem
combines calculate
to form the difference
a unified between
area. If the two geometric
regions have areas.
areas Ba. In It creates
this criterion, the aim is to minimize the degree
common boundaries, these boundaries will be included in the of mutual coverage of the hose ranges of their operation.
resulting region, which is the result of subtracting one region from the other. The
resulting union region. The region difference function is used
to calculate of
hematical description thethe
difference
functionbetween
is astwo geometric
follows: 𝐴𝐴 areas.
= 𝑅𝑅₁It− 𝑅𝑅₂. This means that the
creates the resulting region, which is the result of subtracting
erence of theone 𝑅𝑅₁ and
setsregion from𝑅𝑅₂the other. The
consists mathematical
of those elements 𝑥𝑥 that belong
description of to the set 𝑅𝑅₁ (𝑥𝑥 ∈ 𝑅𝑅₁)
the function is as follows: A = R₁ – R₂. This means that the
do not belong to the set
difference 𝑅𝑅₂ sets
of the (𝑥𝑥 R₁
∉ and
𝑅𝑅₂).R₂Mathematically, this means
consists of those elements x that the resulting area 𝐴𝐴
that belong to the set R1 (x  R₁) but do not belong to the set
qual to the difference between the set of points 𝑅𝑅₁ and the set of points 𝑅𝑅₂. A point belonging
R₂ (x  R2). Mathematically, this means that the resulting
area A is to
₁ and not belonging equal to thealso
𝑅𝑅₂ will difference
belongbetween the set of points
to the resulting area 𝐴𝐴. The goal of the algorithm
R₁ and the set of points R₂. A point belonging to R₁ and not
belonging
find the smallest to R₂
total willAalso
area a ofbelong to the resulting
the zones not covered area A.
byThe
the range of the hoses. Then,
goal of the algorithm is to find the smallest total area Aa of
objective function
the zonestakes
not the following
covered form:of the hoses. Then, the
by the range
objective function takes the following form:
𝑓𝑓1 (𝑥𝑥) = ∑𝑛𝑛𝑎𝑎=1 𝐴𝐴𝑎𝑎 (𝑥𝑥) Fig. 6 Simulation

(3) results for criterion
Fig. 6. Simulation (3
results - )minimization
for of ofoverlap
criterion – minimization overlap of water st
of water streams
re: where: In this case, the objective function takes the following form:
Aa(x) is the area of the region not covered by the reach of In this case, the objective function takes the following form:
𝐴𝐴𝑎𝑎 (𝑥𝑥) is the
thefire
areahose.
of the region not covered by the reach of the fire hose.
Additionally,
itionally, for better for betterthe
visualization, visualization, the custom
custom preview preview was used, which allows
function = ∑𝑛𝑛𝑎𝑎=1 𝐵𝐵𝑎𝑎 (𝑥𝑥)
𝑓𝑓2 (𝑥𝑥) the (4)
function was used, which allows the user to quickly determine
to quickly determine
the areas notthe areas
covered bynot covered
the range by
of the firethe range of
protection the firewhere:
system. protection system.
where:
Ba(x) is the area of the region of overlap of water streams.
𝐵𝐵𝑎𝑎 (𝑥𝑥) is the area of the region of overlap of water streams.
Simulations were carried out by changing the positions of
Simulations were carried outhydrants.
three by changingThe algorithm placedofthe
the positions hydrants
three in suchThe algo
hydrants.
a way that the overlapping areas of the streams were as small
the hydrants in such a way that the
as possible overlapping
(Fig. 6). areas of the streams were as sma
The third criterion, which is the minimization of the
(Fig. 6). power demand of the fire pump, depends on many factors,
such as the length of the fire protection installation, the
The third criterion, which is the minimization of the power demand of the fire pu
number of hydrants required for simultaneous supply, and
on many factors, such as thethe
pressure
lengthof of
thethe
fire fire
extinguishing
protection agent required inthe
installation, the numbe
fire protection installation. A longer pipeline contributes to
required for simultaneous greater pressure
supply, andlosstheinpressure
the fire protection installation.
of the fire This
extinguishing agen
means that to achieve the required flow at the end point of
the fire protection installation. A longer
the installation, pipeline
the fire contributes
pump must generate to greater
higher pressure lo
pressure
protection installation. compared
This means to a that
shorterto pipeline.
achieve The the pressure
requireddrop flowin at
thethe end
pipeline is proportional to the length of the pipeline and
installation,
Fig. 5. Simulation results for criterion – maximization thearea
of the coverage the flow
fire pump mustvelocity.
generate Thehigher
simplified mathematical
pressure comparedequation
to aofshorter p
of extinguished surfaces Darcy-Weisbach defines, among other things, the relationship
g. 5 Simulation results for criterion - maximization of the coverage area of extinguished
surfaces
as follows [22]: 2 2 2
pressure drop in the pipeline is proportional to the lengthxof, xthe, ypipeline
, yDa,p
a z √are
z ,=and
p
the
|xa
−flow
a the |2velocity.
xpgeometric
+ |ya − coordinates
D a = p√|x
p
yp |2 +|za − zpof
|2 hydrant
for a spatial system
"a" and
− xp | + |ya − yp | +|za − zp |
(XYZ),
pump "p",
for a spatial system (XYZ),
a,p a
The simplified mathematical 𝐿𝐿 equation
𝜌𝜌𝜌𝜌² of Darcy-Weisbach defines, among other things, the
ure droprelationship
in the pipeline𝛥𝛥𝛥𝛥 is
=proportional
𝜆𝜆 𝐷𝐷 2 to the length of thesystemwhere: and the flow(velocity.
Da,p is the shortest
where: distance between
pipeline 5 )hydrant "𝑎𝑎" and pump "𝑝𝑝", 𝑛𝑛 is the number of
between
is the pressure drop [Pa], the length of the pipeline in a hydraulic and the pressure drop in
xa , xp , ya , yp za , zp are the geometric coordinates of hydrant "a" and pump "p",
hydrants. xa , xp , ya , yp za , zp are the geometric coordinates of hydrant "a" and pump "p",
simplified mathematical
the system equation of Darcy-Weisbach defines, among
as follows [22]: other things, the
Da,p is the shortest distance between "𝑎𝑎" and pump "𝑝𝑝", 𝑛𝑛 is the numb
is the resistance coefficient dependent, among a) others, on Reynolds D is the numbershortest and b) hydrant
distance between hydrant "𝑎𝑎" and pump "𝑝𝑝", 𝑛𝑛 is the numb
a,p
onship between
between thethelength
length of pipeline
of the the𝛥𝛥𝛥𝛥pipeline in a hydraulic
𝐿𝐿 𝜌𝜌𝜌𝜌²
=in𝜆𝜆a hydraulic system and
system and the pressure
hydrants. a) ( 5 drop
) in b)
is theroughness
ative pressure drop
of [Pa],
the pipe [-], 𝐷𝐷 2 hydrants.
ystem asthe pressure
follows drop in the system as follows [22]:
[22]: a)
a)
b)
b)
ss the
thelength
resistance
where:
of thecoefficient
pipeline [m],dependent, among others, on Reynolds number and
Δ𝑝𝑝 of 𝐿𝐿 𝜌𝜌𝜌𝜌²
ative roughness is the pipe𝛥𝛥𝛥𝛥
thepressure = [Pa],
drop
[-], 𝜆𝜆 𝐷𝐷 2 (5) (5)
is the diameter of the pipeline [m],
𝜆𝜆 is the resistance coefficient dependent, among others, on Reynolds number and
s the lengthwhere: of the pipeline [m],
e:s the density of the fluid
relative roughness
Δp is the pressure drop [Pa],
[kg/m³],
of the pipe [-],
s the diameter of the pipeline the[m],
s the is λthe
is the𝐿𝐿of
Δ𝑝𝑝 velocity resistance
is the
pressure thelength coefficient
fluid
drop of[m/s].
[Pa], dependent,
pipeline [m], among others, on
Reynolds number and relative roughness of the pipe [-],
s the density
L isthat
the𝐷𝐷length
of the
is thefluid [kg/m³],
diameter
of the of the pipeline
pipeline [m], [m],
be𝜆𝜆noted
is the the power
resistance of the pump
coefficient dependent,is directlyamongproportional
others, on to7.Reynolds
Fig. the flowresults
Simulation rate forand
number and
criterion – minimization of the power demand
D is the𝜌𝜌 diameter
is the densityof theof pipeline [m],
the fluid [kg/m³], of the fire pump: a) with two hydrants, b) with one hydrant
s the andvelocity of the fluid
ρ isinversely
the densityproportional
of
[m/s].[kg/m³],
droprelative roughness ofthethefluid
pipe [-],to hydraulic efficiency. This means that there is a
𝑢𝑢 is the velocity
u is the velocity of the fluid of the[m/s].
fluid [m/s].
be noted between
that the the power of the pump is directly Fig. 7 Simulation results forflow
criterion
rate-power
minimization of the power demand of the fire pump:
tionship
𝐿𝐿 is the lengthbeofnoted thelength
pipeline of the[m], installation andproportional
the demandFig. to 7the
for the
Figure 7driving
shows
Simulation and
theand
results results
for of a-- simulation
criterion minimization that
of theillustrates
power demand of the fire pu
It should that the power of the pump is directly proportional Fig. 7toSimulation
the flow rate
a) with two hydrants, b) with one hydrantof the power demand of the fire pu
results for criterion minimization
rop It
andpressure should
inversely be noted
proportionalthat the power
to hydraulic of the pump is directly
efficiency. This the power
means that requirements
there is a a) for
with a
twogiven area
hydrants, b) depending
with one on the
hydrant
pump. The powerdrop of the fire pump can be expressed by the following
This meansequation:that there isa) awith two hydrants, b) with one hydrant
𝐷𝐷 is the
proportional
diameter and
to inversely
ofthethe
flow rateproportional
pipeline and [m],pressuretodrop hydraulic efficiency.
and inversely
Figure number
7 shows theFigure
results of
of hydrants
a simulation installed.
that illustrates the power requirements
7 shows the results of a simulation that illustrates the power requirements for a givenfor a g
proportional
ionshiplinearbetweenrelationshipto hydraulic
the length
between ofefficiency.
thelength
the This
themeans
installation
of andthat there
the
installation is demand
demand
and the forInfor
Figure the
the case
driving
7 shows
the theofresults
driving the
power
powerfourth criterion,
of a simulation i.e., the the
that illustrates minimum
power requirements for a g
area depending on area
the number ofon hydrants installed.
𝜌𝜌 is the
a linear
of relationship
density
thepower
fire
of the
pump. Thebetween
fluid
power ofthe
thelength
[kg/m³], of thecan
fire pump installation andby the
be expressed area
depending
distance
depending
following from
the number
evacuation
on the
equation:
of hydrants installed.
exits, the
number of hydrants algorithm works in
installed.
pump. The of the
the demand for the driving powerfire pump can
of thebe expressed
fire pump.
In The
the by
casethe
powerof following
thea
In similar
the case
fourth equation:
way
of the
criterion,as described
fourth
i.e., in
criterion,
the criterion
minimumi.e., the 3, withfrom
minimum
distance the difference
distance from evacuation
evacuation exits, theexits
𝑄𝑄∙𝛥𝛥𝛥𝛥
𝑢𝑢 is the velocity
of the fire pump of canthe befluid 𝑃𝑃 [m/s].
= (𝜂𝜂∙𝜌𝜌)
expressed by the following equation: that it takes
algorithm works into
in account ( 6the
a similar ) distance
In the case of the fourth criterion, i.e., the minimum
way as describednotinfrom
distance
the 3,
criterion firewith
from evacuation exits
pump,
the difference that it
algorithm works inalgorithm
a similarworks
way inasadescribed
similar wayinascriterion
described3,inwith the 3,
criterion difference that it takes
with the difference that it
𝑄𝑄∙𝛥𝛥𝛥𝛥
but account
into from the the evacuation exit.theThe
distance not from fire objective function
pump, but from then exit. The obje
the evacuation
ould be noted that the power of the pump is
𝑃𝑃 = (𝜂𝜂∙𝜌𝜌) directly proportional
into account takes
the distance
into to
not
account the
the following
from flow
the fire rate
(not
pump, and
6 )from the fire pump, but from the evacuation exit. The obje
the distanceform: but from the evacuation exit. The objective
𝑄𝑄∙𝛥𝛥𝛥𝛥 function then takes the following form:
ure drop and inversely proportional 𝑃𝑃 = (𝜂𝜂∙𝜌𝜌)to hydraulic function then(6)takesfunction
efficiency. then takes
the following
This means form:
that ( 6 ) is
the following
there form:
a
where: 𝑛𝑛
s the power of the pump [W], 𝑓𝑓 (𝑥𝑥) = ∑𝑛𝑛 𝐷𝐷 (𝑥𝑥)
𝑛𝑛 𝑓𝑓4 (𝑥𝑥) = ∑𝑎𝑎=1 𝐷𝐷𝑎𝑎,𝑤𝑤𝑤𝑤 (𝑥𝑥)
where: between the length of the installation and the demand for the driving
relationship 𝑓𝑓4 (𝑥𝑥) =power ∑𝑎𝑎=14𝐷𝐷𝑎𝑎,𝑤𝑤𝑤𝑤 (𝑥𝑥)𝑎𝑎=1 𝑎𝑎,𝑤𝑤𝑤𝑤 (8) (8)
𝑃𝑃 is the power of the pump [W],
P is the power of
is the pump flow rate [m³/s], the pump [W], and:
and:
Q is the
e fire pump. 𝑄𝑄pump
The ispowerflowof
the pump rate
the[m³/s],
flow fire[m³/s],
rate pump can beand: expressed by the and:following equation:
s the power ofhydraulic
the pump [W], of the pump [-]. Da,we = √|xa − xwe |22 + |ya − ywe |22 - for a planar system (XY),
η is the
s the hydraulic efficiency D a,we = √|xa − xwe | + |ya − ywe | – -forfora aplanar system(XY),
planar system (XY),
𝜂𝜂 isefficiency
the hydraulic ofefficiency
the pump [-].pump [-].
of the Da,we = √|xa − xwe | 2 + |y − y |2
Da,we = √|x a a −we xwe |22 + |ya − ywe |22 +|zasystem
- for a planar − zwe |22(XY),
– afor
for a spatial
spatial system (XYZ),
s the pumpThe flow rate to[m³/s], Da,we = √|xa − xwe | + |ya − ywe | +|za − zwe | for a spatial system (XYZ),
on to the solution
Theproposed the the proposed
to optimization optimization
criterion will,criterion will,be to system (XYZ),
solution proposed optimization
𝑄𝑄∙𝛥𝛥𝛥𝛥 Dtherefore,
criterion will, therefore,
a,we = √|xa − xwe
be find
|to
2+
where: findthe
|y theminimum
minimum ofof 2
a − ywe | +|za − zwe | for a spatial system (XYZ),
2
therefore, be to find the minimum 𝑃𝑃 = of the objective function, where:
(6)
sive
thefunction,
hydraulic efficiency
the objective is the of
function, theispump
which [-].
thebetween
distance
(𝜂𝜂∙𝜌𝜌) between each fire hydrant
andand 𝑥𝑥the fire
𝑎𝑎 ,fire
𝑥𝑥𝑤𝑤𝑤𝑤 pump, with
, 𝑦𝑦pump,
𝑎𝑎 , 𝑦𝑦𝑤𝑤𝑤𝑤 𝑧𝑧𝑎𝑎 ,with
𝑧𝑧𝑤𝑤𝑤𝑤 are the geometric coordinates of hydrant "𝑎𝑎" and evacua
which is the distance between each fire hydrant
which distance and the fire
where: each fire hydrant where:𝑥𝑥𝑎𝑎 , 𝑥𝑥𝑤𝑤𝑤𝑤 , 𝑦𝑦𝑎𝑎 , 𝑦𝑦𝑤𝑤𝑤𝑤 𝑧𝑧𝑎𝑎 , 𝑧𝑧𝑤𝑤𝑤𝑤 are the geometric coordinates of hydrant "𝑎𝑎" and evacua
the
sly
apump, with
previously a previously
defined defined
location on thelocation
ship: on the ship:
on to the proposed optimization criterion will, therefore, be, 𝑦𝑦𝑎𝑎to, x𝑦𝑦a𝑤𝑤𝑤𝑤
𝑥𝑥𝑎𝑎 , 𝑥𝑥𝑤𝑤𝑤𝑤
, xwe,exit
find ya, y"𝑤𝑤𝑤𝑤",
,the
𝑧𝑧𝑎𝑎exit
, za, zwe, are the
minimum
"𝑤𝑤𝑤𝑤",
we
𝑧𝑧evacuation
𝑤𝑤𝑤𝑤 are the geometric
of geometric coordinates of hydrant
coordinates of hydrant "𝑎𝑎" and evacuation
e: defined location on the ship: 𝑛𝑛
“a” and exit “we”,
ve function, which is the distance 𝑓𝑓3 (𝑥𝑥) = ∑𝑎𝑎=1
between 𝐷𝐷𝑎𝑎,𝑝𝑝 (𝑥𝑥)
each exit "𝑤𝑤𝑤𝑤",and
fire hydrant D the is fire pump, (with
the shortest 7 )
distance between hydrant “a” and
𝑛𝑛 a,we
𝑃𝑃 is the power of the 𝑓𝑓3 (𝑥𝑥)
pump ∑𝑎𝑎=1 𝐷𝐷𝑎𝑎,𝑝𝑝 (𝑥𝑥)
= [W], (7) evacuation exit “we”. ( 7 )
ly defined
and:location on the ship: a) b)
𝑄𝑄 is the
and:pump flow rate [m³/s],
2 2
Da,p = √|xa − x𝑓𝑓p |(𝑥𝑥)
+ |y=a − yp𝑛𝑛| for a planar
a planarsystem
system(XY),
∑ 𝑎𝑎,𝑝𝑝 (𝑥𝑥)[-].
the𝐷𝐷pump (7)
for (XY),
𝜂𝜂 is the hydraulic efficiency
3 of𝑎𝑎=1
2 2
2 2 2
|x = √|xa−−yxp|| + |yfor
D|a,p + a −a
ypplanar
| +|za − for
zp | for a spatial
a − xpto the|yproposed system a spatial
(XY), system (XYZ),
olution a p optimization criterion will, therefore, be to find the minimum of
system (XYZ),
bjectivewhere:
function,
2
which 2
is the distance between each fire hydrant and the fire pump, with
xa − xpwhere:
| + |y xaa, x− , y
y , |
p ap p a , zfor
y z a the
p are planar system
geometric (XY), of hydrant "a" and pump "p",
coordinates
viously xdefined
a
, xp, ya, ylocation
, z , zp areonthe
p a
thegeometric
ship: coordinates of hydrant
“a” andDpump “p”,shortest distance between hydrant "𝑎𝑎" and pump Fig.
a,p is the "𝑝𝑝", 𝑛𝑛 is the number
8 Simulation of
results for criterion – minimum distance from evacuation
Da,p is the shortest distance between𝑛𝑛 hydrant “a” and pump exits: a) the evacuation exit is on the left side of the room, b) the evacuation
𝑓𝑓 (𝑥𝑥) =∑ 𝐷𝐷 (𝑥𝑥) ( 7 side
) of the room
“p”, n is the number of3hydrants. 𝑎𝑎=1 𝑎𝑎,𝑝𝑝
hydrants. exit is on the right
a) b)
Two simulations were conducted, changing the location
of the evacuation exits (Fig. 8). In Figure 8a, the evacuation
2 2 exit is on the left side of the room, while in Figure 8b, it is on
= √|xa − xp | + |ya − yp | for a planar system (XY), the right. The simulation parameters as well as the number
of hydrants in both cases are the same. The algorithm placed
the hydrants in the vicinity of the evacuation exits, which

vacuation exit is on the left side of the room, while in Figure 8b, it is on the
ation parameters as well as the number of hydrants in both cases are the same.
laced the hydrants in the vicinity of the evacuation exits, which confirms the
tion of placing the hydrants
confirms as assumption
the fulfilled close to them as possible.
of placing the hydrants as
close to them as possible.
hat the designer would like to take into account several of the mentioned
It is obvious that the designer would like to take into
account several
teria simultaneously. of the
In this mentioned
case, optimization
these criteria shouldcriteria
be merged into one
simultaneously. In this case, these criteria should be merged
into one
bjective function. representative
There are manyobjective
known function.
methodsThere are many classical multi-
for reducing
known methods for reducing classical multi-criteria
ation to single-criteria
optimizationones, such as the
to single-criteria hierarchical
ones, optimization method, the
such as the hierarchical
optimization method, the method of constrained criteria,
rained criteria, and the global criterion method.
and the global criterion method.
In ourand
a relatively simple approach,
often useda relatively
weighted simple and method
criteria often usedwas applied. This
weighted criteria method was applied. This method
combines
es two criteria into atwosingle
criteriaobjective
into a single objectiveusing
function functiona using
weighted sum. The results for two criteria – maximizing the coverage of
Fig. 9. Simulation
a weighted sum. The substitute objective function would be extinguished surfaces and minimizing the distance to evacuation exits
tive functiona linear
wouldcombination
be a linear combination
of the of theeach
two criteria, where two criteria, where each
criterion
is multiplied by a weight factor that reflects its relative It was assumed that the most relevant of the four criteria
iplied by a weight factor that reflects its relative importance. The weights can
importance. The weights can be determined by the designer considered was the one that affects the degree of coverage
y the designerbased on recognized
based priorities,
on recognized for example,
priorities, the type of athe
for example, shiptypeof
ofthe room.
a ship orThe simulation showed that the extinguished
or ship-owner preference. In our approach, the total value surfaces were fully covered and the distance from the
erence. In ourof the substitutethe
approach, objective function,
total value of which representsobjective
the substitute the emergency
function, exit also appeared to be optimal.
established criteria, should be minimized:
s the established criteria, should be minimized:
Results and discussion
𝐹𝐹𝑠𝑠𝑠𝑠𝑠𝑠 (𝑥𝑥) = 𝑚𝑚𝑚𝑚𝑚𝑚 ∑𝑘𝑘𝑖𝑖=1 𝑤𝑤𝑖𝑖 ∙ 𝑓𝑓𝑖𝑖 (𝑥𝑥) (9) (9)
As previously mentioned, the main goal of the developed
where: system, i.e., the computer-aided system for the layout of fire
k is the number of objective functions, hydrants on board designed vessels, is to reduce the designer’s
function, fi(x) is the value of the i-th objective function, working time. It is difficult to estimate the working time of
w i are the weights of the objective function such that a designer who places hydrants on a ship’s board using the
ction such that 𝑤𝑤 ϵ [0,1] and ∑𝑘𝑘𝑖𝑖=1 w𝑖𝑖 = 1. classical method. On the one hand, the designer’s working
time depends on their knowledge, skills, and experience, for
As a rule, the substitute objective function can be subject to example expressing itself through routines, such as storing
different criteria
n be subject to different criteria and
and may
may require
require normalization under certain subject-specific regulations in his memory. On the other hand,
circumstances. Normalization is the process of scaling or this time depends on the type of ship and the constraints that
Normalization is the process of scaling or
transforming data to bring it into a specific range or format. exist in it, such as the scope of regulations that depend on
Thisis is
ge or format. This particularly
particularly important
important whenwhen dealing with multiple the type of ship. In addition, designers may also be involved
the objective function has different units or function has different units
objectives or when the objective in other design work.
or magnitudes. This can be done using various techniques, but From the experience of one of the co-authors of this study
ous techniques, but study,
in this in this study, normalization
min-max min-max was applied. A detailed as a lead engineer in a ship design office, the time required to
description
tion of such a method of by
utilized such theaauthors
methodcan utilized by the authors can be complete the arrangement drawing is about 30 to 40 hours of
found in [23]. design work. Assuming that nearly half of the time is taken
Weights are assigned to each of the objective functions, up by the analysis of the placement of hydrants, and the rest
reflecting
functions, reflecting theirtheir relative
relative importance
importance to to the overall evaluation is the pure drawing part, it is safe to say that potentially up
of the solution. Various
methods can be used to determine these weights, methods can be used to determine to a dozen hours of work can be saved.
these weights, such as the analytical hierarchical process To evaluate the effectiveness of the developed system three
HP) method. Once (AHP) method.forOnce
the weights each the weights for each criterion have
criterion different types of ships were considered. On board considered
the criteria for each solution was calculated, sum of the criteria for each
been determined, the weighted ships, hydrants were placed using the classical manual method
solution was calculated, allowing the results to be compared and their layout was compared to the layout proposed by the
sis of a single indicator and ultimately selecting
on the basis of a single indicator and ultimately selecting the developed system. These were ships of different types, namely:
solution with the highest value. –– Fishery Researche Vessel (Fig. 10),
The result of one such two-criteria simulation: maximizing –– Wind Platform Vessel (Fig. 11),
on: maximizing the coverage of extinguished
the coverage of extinguished surfaces and minimizing the –– Multi Role Auxiliay Vessel (Fig. 12).
evacuation exitsdistance to evacuation
is presented on Figureexits
9. is presented on Figure 9. It was assumed that the classical method of hydrant
placement meets all the requirements (constraints) that
were set before the designer. Figures 10, 11, and 12 show
top view projections of selected rooms on the mentioned
ships. In this case, specific details of the designed vessels

were deliberately not provided to protect the intellectual authors’ experiences working in a ship design office, taking
property of ship-owners and the design office. Arrangements into account the type of vessel being designed. The example
of individual technical rooms were modified, but the location values of weights for individual partial objective functions
and distribution of fire protection equipment remained have been presented in the captions of Figures 10, 11, and 12.
unchanged. The placement of fire hydrants obtained using the In particular, when formulating the substitute objective
classical method is marked using markers in the form of red function, the following criteria were taken into account:
circles, whereas the placement obtained using the developed maximizing the coverage of extinguished surfaces and
system was overlaid on each projection using markers in the minimizing the distance to evacuation exits for all considered
form of blue crosses. vessels, i.e., the Fishery Research Vessel (Fig. 10), the Wind
In this case study, the values of individual weights of the Platform Vessel (Fig. 11), and the Multi Role Auxiliary Vessel
objective function were determined based on one of the (Fig. 12).
Fig. 10. Comparison of the actual arrangement of fire hydrants on a fishing vessel with their simulated distribution
based on two selected criteria: maximizing the degree of coverage of extinguished areas and minimizing the distance from evacuation exits
with weights w1 = 0,6 and w4 = 0,4 respectively
Fig. 11 Comparison of the actual arrangement of fire hydrants on a vessel servicing wind platforms
with their simulated distribution based on two selected criteria: maximizing the degree of coverage of extinguished areas and
minimizing the distance from evacuation exits with weights w1 = 0,6 and w4 = 0,4 respectively

Fig. 12. Comparison of the actual arrangement of fire hydrants on a multi-purpose support vessel with their simulated distribution
based on two selected criteria: maximizing the degree of coverage of extinguished areas and minimizing the distance from evacuation exits
with weights w1 = 0,6 and w4 = 0,4 respectively
Based on the comparative analysis of the individual and details, the larger the file size, and consequently, the
solutions regarding the arrangement of fire hydrants on the computer’s computational power required for processing and
decks of various ships, the following can be concluded: displaying graphics in real-time becomes greater.
–– both the layout of the fire hydrants obtained using the Tab. 1. Simulation duration for the Wind Platform Vessel, used for wind
classical method and the one obtained using the developed platform servicing, based on the criteria of maximizing the coverage
system meet all the imposed requirements (constraints), degree of extinguished surfaces and minimizing the distance to
evacuation exits
–– the layout of the fire hydrants obtained using the developed
system enables faster access by the firefighting team to number of hydrants
the hydrants, thanks to the criterion of minimizing the
9 12 15 18 25 50
distance from evacuation exits as a component of the
substitute objective function, which significantly increases Max. speed = 0.1 11 11 11
12[s]
14 15
the level of protection against fire hazards on the ship. Iterations: 20 [s] [s] [s] [s] [s]
Considering that the main goal of the developed system Max. speed = 0.2 11 11 11 14 15
12[s]
is to reduce the workload of the designer, the above- Iterations: 20 [s] [s] [s] [s] [s]
mentioned benefits are an additional but very significant
Max. speed =` 0.3 11 11 12 14 15
element of the feasibility of the developed system. So, the Iterations: 20 [s] [s] [s]
12[s]
[s] [s]
basic question remains – how quickly can a solution related
to the arrangement of fire hydrants be obtained using the Max. speed = 0.5 11 11 11 13 15
12[s]
Iterations: 20 [s] [s] [s] [s] [s]
developed system?
In order to answer the question above, a series of simulations Max. speed = 1 11 11 11
12[s]
14 15
were conducted using a computer with typical parameters Iterations: 20 [s] [s] [s] [s] [s]
for computers used in ship design offices. The results of Max. speed = 0.1 21 22 22 22 24 29
the simulation duration for the Wind Platform Vessel are Iterations: 40 [s] [s] [s] [s] [s] [s]
shown in Table 1. Similar simulations were conducted for the
Max. speed = 0.2 21 21 22 23 24 28
remaining three ships. Using the PSO algorithm, the number Iterations: 40 [s] [s] [s] [s] [s] [s]
of iterations, speed, and number of particles (individuals)
in the search space were varied. The type of analyzed vessel Max. speed = 0.3 21 22 22 23 24 28
Iterations: 40 [s] [s] [s] [s] [s] [s]
seemingly had an impact on the algorithm’s speed, but this
was solely due to the size of the graphic file that served as Max. speed = 0.5 21 21 22 22 24 29
the source of the analyzed unit’s general plan. The size of the Iterations: 40 [s] [s] [s] [s] [s] [s]
graphic file depends on various factors, such as the number Max. speed = 1 21 22 22 22 24 29
of graphic elements, resolution, number of layers, styles, Iterations: 40 [s] [s] [s] [s] [s] [s]
and blocks contained in the file. The more graphic elements

Changing the algorithm parameter of particle speed in References
the search space did not affect the algorithm’s runtime.
Increasing the number of hydrants used in the simulation 1. R. J. Cant and C. S. Langensiepen, “Methods for Automated
(from 9 up to a maximum of 50) significantly increased the Object Placement in Virtual Scenes,” in 11th International
algorithm’s runtime. The swarm size, determined by the Conference on Computer Modelling and Simulation, UKSim
number of particles (individuals) included in it, also had 2009, 2009. doi: 10.1109/UKSIM.2009.69.
a significant impact on the algorithm’s speed. In the extreme
case, the time required to perform several dozen iterations 2. K. Xu, J. Stewart, and E. Fiume, “Constraint-based
was about 3 minutes, which compared to the time spent on automatic placement for scene composition,” in Proceedings
placing hydrants in a classical way, seems to be an undeniable – Graphics Interface, 2002.
advantage of the applied method.
3. R. Z. Farahani, N. Asgari, N. Heidari, M. Hosseininia, and
M. Goh, “Covering problems in facility location: A review,”
Conclusions Computers and Industrial Engineering, vol. 62, no. 1. 2012.
doi: 10.1016/j.cie.2011.08.020.
The proposed method is an attempt to support ship system
designers using one of the methods of artificial intelligence, 4. S. Dantrakul, C. Likasiri, and R. Pongvuthithum, “Applied
namely the PSO algorithm, which simulates the behavior p-median and p-center algorithms for facility location
of a particle swarm moving in a search space for the best problems,” Expert Syst Appl, vol. 41, no. 8, 2014, doi:
solution to a problem, such as placement of fire hydrants 10.1016/j.eswa.2013.11.046.
on a ship’s boards. The most important conclusions can be
formulated as follows: 5. B. Zhang, J. Peng, and S. Li, “Covering location
–– the proposed method reduces the time-consuming process problem of emergency service facilities in an uncertain
of firefighting equipment layout on a ship’s boards while environment,” Appl Math Model, vol. 51, 2017, doi: 10.1016/j.
also easily implementing the general plan of the designed apm.2017.06.043.
vessel in the Rhino environment,
–– the use of metaheuristic optimization algorithms such 6. S. Y. Kim, B. Y. Moon, and S. C. Shin, “Evaluation criterion
as PSO can provide solutions in a relatively short time, of machinery arrangement design in a ship engine room,”
ranging from a few seconds to several minutes, which is Journal of Ship Production, vol. 25, no. 3, 2009, doi: 10.5957/
significantly shorter than the time required for manual jsp.2009.25.3.117.
design of hydrant placement, making the developed system
much more efficient, 7. P. Gualeni, F. Perrera, M. Raimondo, and T. Vairo,
–– as a result of optimal placement of fire hydrants based “Accessibility for maintenance in the engine room:
on selected criteria, additional benefits arise in the form development and application of a prediction tool for
of increased safety (faster access by firefighting teams operational costs estimation,” Ship Technology Research,
to hydrants) and resource economy (reduced amount of 2022, doi: 10.1080/09377255.2021.2020949.
water required to extinguish a fire or decreased electricity
expenditure to power fire pumps). 8. J. K. Lee, K. H. Lee, N. S. Park, Y. U. Jang, J. Y. Bae, and H.
S. Shim, “Integration of knowledge-base and CAD system
Further research for the machinery layout design of ships,” in Proceedings of
The developed system is based on four different criteria the ASME Design Engineering Technical Conference, 1996.
for the optimal placement of fire hydrants; however, its doi: 10.1115/96-DETC/DAC-1070.
further development seems reasonable. Directions for further
research in the field of hydrant placement system using the 9. A. R. Chaudhuri and U. P. Thakur, “Design criteria of piping
PSO algorithm may include: system for ships,” Journal of the Institution of Engineers
–– optimization of the process of finding the minimum (India), Part MR: Marine Engineering Division, vol. 81,
number of hydrants to reduce the cost of the ship’s fire no. 1, 2000.
safety system while ensuring fire safety requirements,
–– supplementing the developed system with additional 10. B. C. Lee, Y. Choi, and H. Chung, “Firefighting equipment
optimization criteria depending on the specificity of the arrangement optimization for an offshore platform
designed vessels, such as installation costs, considering travel distances,” J Mar Sci Eng, vol. 9, no. 5,
–– using other metaheuristic optimization algorithms, such 2021, doi: 10.3390/jmse9050503.
as genetic algorithms or ant colony algorithms, to compare
the effectiveness of different algorithms in the context of 11. X. Liu, B. Sun, Z. D. Xu, and X. Liu, “An adaptive Particle
hydrant placement. Swarm Optimization algorithm for fire source identification
of the utility tunnel fire,” Fire Saf J, vol. 126, 2021, doi:
10.1016/j.firesaf.2021.103486.

12. K. S. Lee, K. J. Han, and J. W. Lee, “Feasibility study on
parametric optimization of daylighting in building shading
design,” Sustainability (Switzerland), vol. 8, no. 12, 2016,
doi: 10.3390/su8121220.
13. D. Rutten, “Galapagos: On the logic and limitations of

generic solvers,” Architectural Design, vol. 83, no. 2, 2013,
doi: 10.1002/ad.1568.
14. C. Waibel, T. Wortmann, R. Evins, and J. Carmeliet,

“Building energy optimization: An extensive benchmark
of global search algorithms,” Energy Build, vol. 187, 2019,
doi: 10.1016/j.enbuild.2019.01.048.
15. J. M. Cichocka, A. Migalska, W. N. Browne, and E. Rodriguez,

“SILVEREYE – The implementation of particle swarm
optimization algorithm in a design optimization tool,” in
Communications in Computer and Information Science,
2017. doi: 10.1007/978-981-10-5197-5_9.
16. T. Wortmann, “Opossum – Introducing and Evaluating

a Model-based Optimization Tool for Grasshopper,” in
Proceedings of the 22nd Conference on Computer Aided
Architectural Design Research in Asia (CAADRIA), 2022.
doi: 10.52842/conf.caadria.2017.283.
17. I. Anton and D. TĂnase, “Informed Geometries. Parametric

Modelling and Energy Analysis in Early Stages of Design,”
in Energy Procedia, 2016. doi: 10.1016/j.egypro.2015.12.269.
18. J. C. Lagarias, J. A. Reeds, M. H. Wright, and P. E. Wright,

“Convergence properties of the Nelder-Mead simplex
method in low dimensions,” SIAM Journal on Optimization,
vol. 9, no. 1, 1998, doi: 10.1137/S1052623496303470.
19. J. Kennedy and R. Eberhart, “Particle Swarm Optimization,

Proceedings of IEEE International Conference on Neural
Networks Vol. IV: 1942–1948.,” Neural Networks.
20. L. M. Hiot et al., Handbook of Swarm Intelligence, vol. 8.

2010.
21. S. Bhattacharyya and P. Dutta, Handbook of research

on swarm intelligence in engineering. 2015. doi:
10.4018/978-1-4666-8291-7.
22. G. O. Brown, “The history of the Darcy-Weisbach

equation for pipe flow resistance,” in Proceedings of the
Environmental and Water Resources History, 2002. doi:
10.1061/40650(2003)4.
23. K. Rudzki, P. Gomulka, and A. T. Hoang, “Optimization

Model to Manage Ship Fuel Consumption and Navigation
Time,” Polish Maritime Research, vol. 29, no. 3, 2022, doi:
10.2478/pomr-2022-0034.

10.2478/pomr-2023-0054
RETROFITTING THE BOW OF A GENERAL CARGO VESSEL

ANDEVALUATING ENERGY EFFICIENCY OPERATIONAL INDEX
Yordan Denev
Technical University of Varna, Bulgaria
* Corresponding author: y.denev@tu-varna.bg (Yordan Denev)
Abstract
This report examines the feasibility and impact of retrofitting the bulbous bow on a general cargo ship, in terms of the
energy efficiency operational index (EEOI), in the areas of Western Europe and the Eastern Mediterranean. Three
ship forms were developed and analysed: with a bulbous bow, without a bulbous bow, and with a modified bulbous
bow. The goal in developing the ship forms and conducting the analysis was to achieve minimal differences in the ship’s
characteristics with the same volumetric displacement, aided by PolyCAD software. A route was selected between
two ports: Varna and Rotterdam. The labour intensity of the bulbous bow retrofitting process was evaluated and
approximate values of labours costs and cost for the task were determined. The results obtained for resistance during
ship motion, EEOI, and fuel consumption reductions, or increases, were compared against the retrofitting values. The
return cost of retrofitting is evaluated and measured in terms of fuel saved.
Keywords: retrofitting, bulbous bow, energy efficiency design index, return costs
introduction gas, modifying the bow design, or simply changing the

configuration of the bulbous bow.
In recent years, the International Maritime Organization Retrofitting bulbous bows is an interesting process, not
(IMO) introduced requirements for reducing emissions to the only for large container vessels, but also for other types
environment. The IMO strategy aims to reduce freight rate of merchant ships. Force technology is carried out in the
carbon emissions by 40% by 2030, compared with 2008, and retrofitting of the bulbous bow of multipurpose vessels with
up to 70% by 2050 [1]. On January 1 2023, new regulations 9100 DWT. The result is a 17.5% resistance saving [3].
took effect, relating a ship’s Energy Efficiency Existing Ship A numerical analysis of retrofitting a bulbous bow for
Index (EEXI) calculations to energy efficiency and initiating a modern container ship, operating with a slow-steaming
the annual operational carbon intensity indicator (CII) [2]. profile, was presented in [4]. The retrofit analysis served as an
Taking the International Maritime Organization illustrative example of a design process that relies on high-
regulations into account, ship owners are required to take fidelity CFD simulations and surrogate modeling. The bulbous
measures to improve the energy efficiency of their fleet as bow design candidates were generated by parametrically
part of the global effort. Some of the measures to improve modifying the original bow geometry. These alternative
the energy efficiency of ships consist of methods to reduce designs were assessed using the open-source CFD toolbox
the ship’s resistance, use renewable energy, converting Open FOAM, and the resulting effective power predictions
conventional diesel engines to operate on liquefied natural were used to rank each design throughout the entire operating

profile. Moreover, the impacts of the different bulbous bow METHODOLOGY AND SECTION
designs on wave-making resistance and propeller performance MODELLING
were thoroughly examined. Surrogate models were then
employed to explore the parameterized design space and The methodology and section modelling are related to
establish a sequence of design exploration and exploitation modernisation process descriptions in different stages of
cycles in the retrofit analysis, aiming to achieve an enhanced the calculations.
bow shape as the ultimate objective.
An assessment of the design and operational energy Energy Efficiency Operational Indicator
efficiency index of a group of container ships from Class A13, Evaluation
A15, and A19 was conducted in [5]. It transpired that, based
on these indicators, the best performer was the container The IMO guidelines [10] define the methodology for the
ship from Class A19, while Class A13 would need to reduce calculation of the Energy Efficiency Operational Indicator
its speed by 45% to meet the requirements for lowering index (EEOI) based on voyage parameters and the type of main
values. The possibility of using a liquefied natural gas engine engine fuel. EEOI is defined individually for cargo ships and
for the Class A19 ship could enhance its energy efficiency, bulk carriers defined in a wide individually deadweight forrange cargo [11]ships but itand can be bulk carr
defined
resulting in savings of approximately $27 million individually [5]. for cargo calculated ships calculated by the following equa�on [1]: range
by and
the bulk
following carriersequation in a[1]: wide deadweight
A possible solution for reducingcalculated
harmful emissions by the following into equa�on [1]:
the atmosphere is a hybrid propulsion system. Applied to
defined individually for cargo ships and=bulk
𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸 carriers in
∑ 𝑭𝑭𝑭𝑭𝑭𝑭𝑭𝑭∗𝑭𝑭𝑭𝑭𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪
, a wide
𝒕𝒕𝒕𝒕𝑭𝑭𝑭𝑭𝑪𝑪𝑪𝑪𝟐𝟐𝟐𝟐 deadweight rang
a container ship of Class A19, respective reductions of NOx,
∑ 𝑭𝑭𝑭𝑭𝑭𝑭𝑭𝑭∗𝑭𝑭𝑭𝑭𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪 𝒕𝒕𝒕𝒕𝑭𝑭𝑭𝑭𝑪𝑪𝑪𝑪𝟐𝟐𝟐𝟐 (1)
SOx, and CO2, by 52.0%, 63.7%, and 30.4%, calculated
𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸were
= achieved, by the following
, equa�on [1]: ∑ 𝒎𝒎𝒎𝒎𝒎𝒎𝒎𝒎∗𝑫𝑫𝑫𝑫 𝒕𝒕𝒕𝒕𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪
∑ 𝒎𝒎𝒎𝒎𝒎𝒎𝒎𝒎∗𝑫𝑫𝑫𝑫 𝒕𝒕𝒕𝒕𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪
compared to a conventional system. Additionally, it is a more where Fc is the specific fuel consumption (g/kWh), Ccarbon
efficient option concerning environmental regulations, ∑ 𝑭𝑭𝑭𝑭𝑭𝑭𝑭𝑭∗𝑭𝑭𝑭𝑭𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪 where
𝒕𝒕𝒕𝒕𝑭𝑭𝑭𝑭𝑪𝑪𝑪𝑪 toFc is massthe specific fuel consump�on for fuel, mi is(g/kWh
where Fc is𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸𝐸
the =with∑ fuel
specific
is the fuel , mass
consump�on
𝟐𝟐𝟐𝟐 CO
(g/kWh),
conversion
Ccarbon
factor
is the fuel(t), mass andtoDCO2
2
an energy cost of $0.07/kWh and profitability of $21.9/ton [6]. 𝒎𝒎𝒎𝒎𝒎𝒎𝒎𝒎 the cargofactor
∗ 𝑫𝑫𝑫𝑫 carriedfor
𝒕𝒕𝒕𝒕𝑪𝑪𝑪𝑪𝑪𝑪𝑪𝑪 (t),fuel, and Dmi is the is the distance cargoincarried nautical miles is t
factor bulb
The implementation of a double-hull for fuel, on the mibow is the of cargo to thecarried cargo (t),
work done (nm).
carried and or D
work is the
done distance
(nm). in nau�cal miles to th
work
a fishing vessel with a non-optimized hull donewhere
directly (nm). Fc is the specificProper
impacted fuel consump�on evaluation of (g/kWh),
the retrofitting Ccarbon effect is the
on the fuel mass to C
energy
Proper evalua�on of the retrofi�ng effect o
the factor
its operational efficiency. Following Proper modernization for fuel,of
evalua�on miofis efficiency the
the cargo retrofi�ng carried effect
operational (t), index
andon Dhas isthe the distancethe
to calculate
energy in nau�cal
efficiencyfuel costopera�o miles to
the shape and towing tests, a reductionwork in resistance of before and calculate
after the fuel
retrofitting. Fuelcost cost before is calculated and a�er by: retrofi�ng
approximately 10% was observed [7].
calculate thedone fuel (nm). cost before and a�er retrofi�ng. Fuel cost is calculated by:
Proper evalua�on of the retrofi�ng effect on the energy efficiency opera
An evaluation of the resistance of a tanker during beam
calculate the fuel cost before and𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐
a�er= retrofi�ng. 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐 by:
= 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹
seas was conducted in [8]. Through simulation, the maximum 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 ∗ 𝑉𝑉𝑉𝑉𝑐𝑐𝑐𝑐 ∗Fuel cost∗is𝐷𝐷𝐷𝐷,
𝐹𝐹𝐹𝐹𝐶𝐶𝐶𝐶𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 calculated
$ (2)
and minimum wave angles were identified, at which the
additional resistance reached its where
maximum and minimum where
where
Fc is the
Fc 𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐𝑐
𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹
specific
is the =specific
fuel 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 ∗ 𝑉𝑉𝑉𝑉𝑐𝑐𝑐𝑐fuel
consumption
consump�on
∗ 𝐹𝐹𝐹𝐹𝐶𝐶𝐶𝐶𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹
(l/h),∗ Vs 𝐷𝐷𝐷𝐷, $is the(l/h), Vs
Fc is the specific fuel consump�on (l/h), Vsdistance
is the ship service speed (kn), Cf
values. These were 180 and 150 degrees for the maximum ship service speed (kn), Cfuel is the fuel per $/l, and miles
$/l, and D is the in nau�cal D is theto cargo
$/l, andangle, D is the distance in nau�cal miles to cargo carried or work done (nm).
angle and 130 degrees for the minimum wheretaking Fc is the intospecific distance fuelinconsump�on nautical miles (l/h),
to cargo Vscarried is theorship work service
done (nm). speed (kn),
account that the degrees of freedom also directly $/l, andinfluenced D is the distance in nau�cal Weight of miles shiptohull cargo carried or work done (nm).
this effect. Weight of ship hull Weight of ship hull
In [9], a new type of bulbous form for ships, with a Froude The weight of the ship’s hull is determined at
TheWeight
number ranging from 0.4 to 0.5, was introduced, weight ofofship
significantly thehull ship’s The hull weight isbaseddetermined of
onthe working at the
ship’s hull
drawingsconstruc�on
is determined and construc�on design at the stagemod and
differing from the conventional ones. This bulb
based on working shape reduceddrawings construction design stage and modernisation is based on
The weight of the and ship’s construc�on
hullfollowed
the is determined models.
expression: at Mathema�cally,
the construc�on the design weight stagecaa
wave generation at high speeds but was sensitive to precise working drawings and construction models. Mathematically,
the followed based expression:
on working
mounting position and velocity. The tests were conducted in drawings the weight and canconstruc�on
be explained models. by the followed Mathema�cally, expression: the weight
the
a towing tank and software simulations were undertaken for followed expression: 𝑊𝑊𝑊𝑊ℎ𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 = 𝑊𝑊𝑊𝑊𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 ℎ𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 + 𝑊𝑊𝑊𝑊𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 +
a ship with a Froude number of about 0.45. 𝑊𝑊𝑊𝑊ℎ𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 = 𝑊𝑊𝑊𝑊𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚 ℎ𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 + 𝑊𝑊𝑊𝑊𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵 + ⋯ … … … … … . 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊
The aim of this study was to analyze the economic effect of 𝑊𝑊𝑊𝑊ℎ𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 =where 𝑊𝑊𝑊𝑊𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚𝑚Wmain
ℎ𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 + 𝑊𝑊𝑊𝑊 𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵𝐵is+the
hull ⋯… … … …of…the
weight . 𝑊𝑊𝑊𝑊𝑊𝑊𝑊𝑊ship’s (3)
main
retrofitting a ship’s hull to improve the EEOI. Alongside
where Wmain hull is the weight ofinthe all ship’s main hull (t),(t), WBHD is the weight of tra
the efforts to improve the energy efficiencywhere of existing ships, where Wmainofhull the ship’s isship’s main
the weight hull of the(t), and
ship’s Wimain is all
hull of (t), othe
the
Wmain
in the ship’s main hull (t), WBHD hull is the
and Wiisisthe weight allweight the
of theofother main construc�ons in the ship’s hullof(tt
hull WBHD is the weight
an economic analysis of the benefits must also be carried transverse bulkheads in the ship’s
in the ship’s main hull (t), and Wi is all of the other construc�ons in the ship’s hul
out, since such actions are costly and time-consuming, main hullModernisa�on (t), and Wi is all of cost thecalcula�on
other constructions in the
Modernisa�on
because the ship is not in operation. Retrofitting thecost bowcalcula�on
of ship’s hull (t).
a ship is directly linked to the ship’s stay in Modernisa�on
dry dock. Fromcost calcula�on Modernisa�on costs include the cost of billable
a technological perspective, such a task Modernisa�on
may not be overly costs include Modernisation the billable cost ofhours billable
cost hours
forcalculation
fabrica�ng for retrofi�ng and hullpart fab
complex, but the economic analysis is more Modernisa�on
challenging. costs include the cost of billable hours forand retrofi�ng retrofi�ng and hull f
billable hours for fabrica�ng andequa�on: retrofi�ng parts of the ship’s hull are calculate
Therefore, the article presents and analyses billable
the benefits hoursand for fabrica�ng Modernisation and retrofi�ng costs include parts the cost of the of billable ship’shours hull are for calcula
equa�on:
return on investment in retrofitting theequa�on: bow of a general retrofitting and hull fabrication costs. The billable hours
cargo ship. for fabricating and retrofitting parts of the ship’s hull are𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 = 𝑊𝑊𝑊𝑊
calculated with the𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 simple = 𝑊𝑊𝑊𝑊=
equation:
𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 𝑢𝑢𝑢𝑢 ∗ 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀
ℎ𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑊𝑊𝑊𝑊 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 , 𝑚𝑚𝑚𝑚𝑚, 𝑚𝑚𝑚𝑚𝑚
∗ 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑢𝑢𝑢𝑢
ℎ𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑢𝑢𝑢𝑢
where Whull is the weight of the ship’s hull (t) a
where where
Whull Whull
is the is
weight of theconstruc�on
the weight ofship’s hull (t)
the ship’s hulland MHsteel
(t) and
(mh/t). are are
MHsteel thethe
billable hour
billable ho
18
construc�on (mh/t).
construc�on
POLISH MARITIME RESEARCH, No 4/2023 (mh/t).
Hull fabrica�on costs closely depend on the st
Hull fabrica�on costs closely
Hull fabrica�on depend
costs closely on the
depend
see Eq. (5). on steel priceprice
the steel andand
the the
weight of mod
weight of m
see Eq.see
(5).Eq. (5).
ula�on
include the cost of billable hours for retrofi�ng and hull fabrica�on costs. The
a�ng and retrofi�ng parts of the ship’s hull are calculated with the simple
include the cost of billable hours for retrofi�ng and hull fabrica�on costs. The
a�ng and retrofi�ng parts of the ship’s hull are calculated with the simple
𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 = 𝑊𝑊𝑊𝑊ℎ𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 ∗ 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑢𝑢𝑢𝑢 , 𝑚𝑚𝑚𝑚𝑚 (4)
𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀 = 𝑊𝑊𝑊𝑊 𝑢𝑢𝑢𝑢 ∗ 𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑀𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑢𝑢𝑢𝑢 , 𝑚𝑚𝑚𝑚𝑚 (4)
ght of the ship’s hull (t) andℎ𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢MHsteel
(4)
are the billable hours per ton of steel
where
ght of the Whullhull
ship’s is the
(t)weight of the ship’sare
and MHsteel hullthe
(t) and MHsteel
billable hours per ton of steel
are the billable hours per ton of steel construction (mh/t).
s Hull fabrication
closely depend on thecosts closely
steel pricedepend on the
and the steel price
weight and
of modernised hull parts,
the weight of modernised hull parts, see Eq. (5).
s closely depend on the steel price and the weight of modernised hull parts,
𝑀𝑀𝑀𝑀𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹 = 𝑊𝑊𝑊𝑊ℎ𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 ∗ 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹, $ (5) (5)
where HFC is the𝑀𝑀𝑀𝑀𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹

hull= 𝑊𝑊𝑊𝑊ℎ𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢𝑢 ∗ 𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹,
modernisation cost ($), $ Whull is the (5)
weight of the ship’s hull (t), and Csteel is the final steel price
in the country ($/t). Fig. 1. Diagram of mesh independence
Hull modelling
The hull form was generated by PolyCAD software. To HULL FORM GENERATION
assess the effect of the modification of the ship’s bow, the
resistance was calculated using the Holtrop and Mennen The main dimensions of the ship were L=120.62 m,
method, for speeds ranging from 0-17 knots. The advantage B= 16.00 m, D= 9.03 m, and d= 6.67 m. The ship had a
of the software is the possibility of recalculating ship double bottom and double side, single deck and one hold,
characteristics in the event of some form of change. After with a 116 TEU container capacity. The hold length was
retrofitting, a small difference in mass displacement appeared. about 84.5m, with a double sided width of 1.3 m per side
and a maximum hold breadth of 13.4 m. The service speed
was 15 kn and the main engine type was a ‘5S35ME’ with
MODEL VALIDATION AND VERIFICATION main engine power of 4350 kW. Three different forms with
similar hull coefficients are shown in Table 1. The original
In this type of analysis, it is important for the calculations hull form was without a bulbous bow (VAR1), while the other
to be within a range of 5% tolerance, which is assumed for two had bulbous bows, where the dimensions of the bulb
engineering calculations. Otherwise, if there is more than were different.
5% tolerance, the impact on the characteristics is significant. Tab. 1. Hull form coefficients
The retrofitting process consisted of modernising the forward
ship hull’s form without making changes to the ship’s main VAR1 VAR2 VAR3
dimensions. Cb 0.78 0.76 0.76
The change in the geometry of the bow was achieved by Cp 0.78 0.77 0.76
mounting a bulb with a specific geometry that corresponds Cm 0.99 0.99 0.99
to the original ship’s form. In this case, in order to ensure a Cw 0.90 0.89 0.87
constant displacement of water, changes were made to the
Δ, m3 10518.34 10518.64 10518.80
coefficients of the shape, specifically the prismatic coefficient
(Cp) and, consequently, the block coefficient (Cb). The
coefficient of the mid-ship section remained the same for all The analysed ship was a general cargo ship with one hold
shapes. Differences in Cb were within 2.5%, and differences of 7000 tDW. The location of the collision bulkhead was at
in Cp were in the range 0.7-2.5%; higher differences in Cw 7% of Lpp, while the engine room bulkhead was at 23% Lpp,
coefficients were in the range 1.8-3.5%. The maximum see Fig. 2 to Fig. 4.
difference in mass displacement was 0.46 t.
In the evaluation of the model shape, a mesh with
rectangular and triangular elements was used. Each type
of element was used in different areas of the ship’s hull. For
example, in the bow and stern regions where the hull shape
has complex curvature in two directions, triangular mesh
elements were used, while rectangular elements were used
in the remaining areas. The transition elements between
rectangular and triangular elements were rhomboidal.
The grid spacing was 0.2 m with a key nudge of 0.002 m; the
number of elements is shown in Fig. 1.
Fig. 2. Variant one (original hull form) without bulbous bow

VOYAGE PARAMETERS
Voyage parameters were selected in accordance with

shipping trends and the transportation of goods between
the Black Sea and Western European ports. The distance
from the port of Varna to the port of Rotterdam is 3940 nm,
as shown in Fig. 6.
Fig. 3. Hull form variant VAR2 with bulbous bow
Fig. 6. Voyage distance map

Fig. 4. Hull form variant VAR3 with modified bulbous bow
The distance of 3940 nm was travelled in 11 days, at an

Retrofitting the forward part of the ship’s hull with operational speed of 15 kn in good weather conditions. The
a bulbous bow decreases the total resistance by about 18% for ship’s main engine was a ‘5S35ME’ type, with the specific fuel
a service speed of 15 knots (Fig. 5), decreasing the necessary consumption and fuel costs shown in Fig. 7. The maximum
main engine power and reducing carbon dioxide in the consumption occurred with the first hull form, which is
atmosphere. without a bulbous bow, and the minimum occurred with
a modified bulbous bow.
Fig. 7. Voyage fuel cost and specific fuel consumption

Fig. 5. Total resistance of ship hulls
Considering the main engine-specific consumption, ship

After the established positive effect of the modification of service speed, distance between ports, and marine diesel
the ship’s bow, it was necessary to determine whether, and to prices, the fuel cost for one voyage was calculated using the
what extent, it led to an improvement in the energy efficiency CEAS engine calculator (specific for different engine powers)
index of the ship, as well as the return on investment, in and is presented in Fig. 6. Average marine diesel oil costs
terms of resources and time invested in the modification for 0.42 €/liter, which corresponds to 546.5 $/mt of very low
regular voyage distances. sulfur fuel oil (according to prices from July 19, 2023) [12].

After retrofitting the forward part of the ship hull, EEOI is improved, which leads to a reduction
CO2 emissions. The reduction measured (as a percentage) is about 4% for ship hull variant VAR2 a
about 3% for hull variant VAR3. The retrofitting effect is not so high but, related to the ship dimensio
it is satisfactory.
Despite the fact that the specific fuel consumption for CAPITAL
the EXPENDITURE
Ship steel hull weight FOR RETROFITTING
estimation ANDbyRETURN
can be evaluated the COST
second variant of the ship’s hull form is the lowest, it does mathematical equations presented in [14] but they are not
Fig.7. Voyage fuel cost and specific fuel consumption
not result in the lowest overall fuel cost when evaluating The effects of appropriate
the in this case,
forward part retrofitting willbecause hull
be clearer weight
after is calculated
calculating capital expenditure, retu
total fuel expenses. This is due to the fact that the difference in relation to weight displacement. In the case
costs, and time for return costs. To study this effect, it is necessary to calculatestudy, the
the hull steel weig
in specific
Considering the mainfuel consumption
engine-specific between
consumption, shipthe second
service and
speed, thirdbetween
distance
billable hours volumetric
for its fabrication,and
ports, weight
and their displacement
differences are forms.
for different the same
Afterfor all
production calculation
marine diesel prices, the fuel cost for one voyage was calculated using the CEAS engine calculator
variants is only 0.8 g/kWh in favor of the second one,
the which
return cost and forms,
time are and computer model development is used for hull
calculated.
cific for different engine powers) and is presented in Fig.6. Average marine diesel oil costs 0.42
does not make
er, which corresponds to 546.5a $/mt
significant
of very low impact
sulfuron fuelthe
oilend result,toasprices
(according the fromweight evaluation. The results are shown in Table 2.
July 19,estimation
Ship steel hull weight can be evaluated by the mathematical equations presented in [1
resistance that needs to be overcome with the second hull Tab. 2. Ship hull
) [12]. but they are not appropriate inweights
this case, because hull weight is calculated in relation to weig
form is higher than the third one.
Despite the fact that the specific fuel consumption for the second displacement. In the
variant of the ship's case
hull formstudy, the volumetric and weight displacement
VAR1 VAR2 areVAR3
the same for all forms, a
e lowest, it does not result in the lowest overall fuel cost when evaluating the total
computer fueldevelopment
model expenses. is used for hull weight evaluation. The results are shown in Table 2.
WEIGHT HULL, T 835.1 852.0 894.0
is due to the fact that the difference in specific fuel consumption between the second and third
ENERGY
nts is only 0.8 g/kWh EFFICIENCY
in favor of OPERATIONAL
the second one, which does not make a significant impactBulb area, m 2
on the
Table 2. Ship hull weights
0.00 10.30 9.03
INDEX EVALUATION BASED ON VOYAGE
esult, as the resistance that needs to be overcome with the second hull form is higher than the third
Bulb length, m 0.00 VAR1 2.00
VAR2 VAR3 3.00
PARAMETERS 852.0
1.50 894.02.00 WEIGHT
Bulb radius, m HULL, T0.00 835.1
2
Bulb area, m 0.00 10.30 9.03
EEOI is an indicator for evaluating ship energy efficiency Bulb length, m 0.00 2.00 3.00
ENERGY EFFICIENCY OPERATIONAL INDEX EVALUATION BASED Bulb radius, m
and CO emissions to the environment during a ship’s
2
The difference in hull weights,
0.00 after1.50
retrofitting,
2.00 is about
ON VOYAGE PARAMETERS
operation and through her life cycle. Using the equation 17 t of steel construction for variant VAR2 and 59 t of steel
for EEOI offered in [10], for generated ship hulls,The indexes
difference construction
in hull weights,forafter
variant VAR3. is
retrofitting, Differences of such
about 17 t of magnitude for variant VAR
steel construction
are calculated considering voyage parameters and
and 59 actual
t of steel do not
construction
EEOI is an indicator for evaluating ship energy efficiency and CO2 emissions to the environment affect
for the ship’s
variant carrying
VAR3. capacity
Differences of since,
such during
magnitude thenot affect the shi
do
ship conditions. For a voyage from the port of Varna
carrying to the
capacity
ng a ship’s operation and through her life cycle. Using the equation for EEOI offered in [10], for conceptual
since, during design
the stages,
conceptual a 1%
design reserve
stages, a 1%displacement
reserve is
displacement is provide
port of
rated ship hulls, Rotterdam
indexes with considering
are calculated a speed ofvoyage
15 kn,parameters
a distance
theandofactual
about
results are shown
ship provided;
in Fig.9.the results are shown in Fig. 9.
conditions.
3940
a voyage from thenm, a Varna
port of deadweight
to the portofofabout 7000
Rotterdam t, and
with a fuel
a speed of 15 carbon
kn, a distance of about
content of
nm, a deadweight ofabout
0.86 7000
for light
t, andfuel oilcarbon
a fuel [13], the EEOI
content calculated
of 0.86 fuel oil [13],198000
for light by the
I calculated by Eq. (1) is presented in Fig.8. 196000
Eq. (1) is presented in Fig. 8.
Hull fabrication, mh/t
194000
5,22E-06
192000
5,20E-06 190000
5,18E-06 188000
EEOI, tCO2/tnm
5,16E-06 186000
5,14E-06
184000
5,12E-06
182000
5,10E-06
0 1 2 3 4
5,08E-06
Variant number
5,06E-06
0 1 2 3 4
Variant number Fig. Fig.9.
9. Billable
Billablehours forsteel
hours for steelhull
hull fabrication
fabrication
Fig. 8. Energy Efficiency

Fig.8. Energy Operational
Efficiency Index
Operational Index for forms
for hull hull forms
The thickness of hull plating in all variants is 16 mm,
which, according to [15], is necessary for 220 mh per ton
After retrofitting the forward part of the ship hull, EEOI of steel construction. This means that, with increasing hull
is improved, which leads to a reduction of CO2 emissions. weight, the billable-hours for fabrication and costs increase
The reduction measured (as a percentage) is about 4% for too. The Chinese steel price is 500 $/t [16], while the USA
ship hull variant VAR2 and about 3% for hull variant VAR3. steel price is 680 $/t [17]. The steel price in Bulgaria is about
The retrofitting effect is not so high but, related to the ship 450$/t. For the purpose of this study, the price of steel in
dimensions, it is satisfactory. Bulgaria is averaged to 600 $/t, including the cost of work,
cost of transportation, cost of blasting and painting, and the
cost of cutting in Bulgaria (Table 3).
CAPITAL EXPENDITURE FOR Tab. 3. Hull cost for different variants
RETROFITTING AND RETURN COSTS VAR1 VAR2 VAR3
The effects of forward part retrofitting will be clearer after Hull Fabrication Cost, $ 501,060 511,200 536,400
calculating capital expenditure, return costs, and time for
return costs. To study this effect, it is necessary to calculate The difference in hull cost price between variant VAR2 and
the hull steel weight, billable hours for its fabrication, and their variant VAR1 (the original) is about $10,000, which is about
differences for different forms. After production calculations, €9000, and equal to about 41 t of very low sulphur fuel oil. For
the return cost and time are calculated. a voyage from the port of Varna to the port of Rotterdam, the
fuel price is about €60,000, which means that, after forward

hull part retrofitting and EEOI improvement, the return https://www.imo.org/en/MediaCentre/PressBriefings/
cost is very fast, i.e. after the second voyage after retrofitting. pages/CII-and-EEXI-entry-into-force.aspx
3. F. Technology, ‘Retrofitting a new bulbous bow’ [online,

CONCLUSIONS viewed 02.07.2023] available https://forcetechnology.com/
da/cases/retrofitting-af-en-ny-bulb
This paper studied the possibilities and effects of retrofitting
the forward part of a ship’s hull designed without a bulbous 4. G. Filip, H. K. Dae, S. Sahu, J. de Kat, and M. Kevin,
bow. To evaluate these effects, three ship hull forms were ‘Bulbous Bow Retrofit of a Containership Using an Open
developed. The original form was without a bulbous bow, Source Computational Fluid Dynamics (CFD) Toolbox.’
while the other two had a bulbous bow. Paper presented at the SNAME Maritime Convention,
Calculations of resistance and engine power were carried Houston, Texas, USA, October 2014. doi: https://doi.
out using the Holtrop and Mennen method and the hull form org/10.5957/SMC-2014-T58
was generated by PolyCAD software. In new form generation,
the volumetric displacement of the original ship hull was 5. A.R. Nader and S.S. Ibrahim, ‘Enhancing energy efficiency
preserved for the newly generated forms. There were small for new generations of containerized shipping’, Ocean
changes in the block and prismatic coefficients. Engineering, Volume 215, 2020, 107887, ISSN 0029-8018,
After forward hull part retrofitting, the total resistance https://doi.org/10.1016/j.oceaneng.2020.107887.
was reduced by about 18%, which lead to a necessary engine
power reduction. To study the numerical retrofitting effects, 6. A.R. Nader and S.S. Ibrahim, ‘Hybrid/dual fuel propulsion
a ship voyage between the port of Varna and the port of systems towards decarbonization: Case study container
Rotterdam was selected, with a distance of 3940 nm. The ship Ocean Engineering, Volume 281, 2023, 114962, ISSN
specific fuel consumption for all designed forms were found 0029-8018, https://doi.org/10.1016/j.oceaneng.2023.114962.
and it should be noted that variant VAR3 has a minimum
specific fuel consumption, but the fuel cost for the voyage is 7. H.R. Díaz-Ojeda, F. Pérez-Arribas, and T.R. Stephen,
not minimal because the necessary power is higher. ‘The influence of dihedral bulbous bows on the resistance
Hull shape variant VAR3 is optimal for EEOI but not of small fishing vessels: A numerical study’, Ocean
optimal for retrofitting or building costs. The capex cost is Engineering, Volume 281, 2023, 114661, ISSN 0029-8018,
about $35,000 higher than variant VAR1 (the original) and https://doi.org/10.1016/j.oceaneng.2023.114661.
about $25,000 higher than variant VAR2. The difference in
fuel cost between variant VAR3 and variant VAR2 is about 8. H.J. Cho, S.H. Lee, D. Oh, and J.K.J. Paik, ‘A numerical
$3800 per voyage, while the difference between VAR1 and study on the added resistance and motion of a ship in
VAR3 is $16,500, and between VAR1 and VAR2 it is about bow quartering waves using a soft spring system’, Ocean
$20,000. Engineering, Volume 280, 2023, 114620, ISSN 0029-8018,
Retrofitting of the forward part of the ship hull is https://doi.org/10.1016/j.oceaneng.2023.114620.
more reasonably carried out using variant VAR3, with a
modified bulbous bow, so that the EEOI is at its lowest and 9. Z. Liu, W. Liu, Q. Chen, F. Luo, and S. Zhai, ‘Resistance
approximately 35 t of very low sulphur fuel oil is saved. Capex reduction technology research of high speed ships based
costs are higher and equal to 64 t of very low sulphur fuel oil, on a new type of bow appendage’, Ocean Engineering,
which means that, after the second voyages, the capex cost Volume 206, 2020, 107246, ISSN 0029-8018, https://doi.
will be returned. org/10.1016/j.oceaneng.2020.107246.
DECLARATION OF COMPETING INTERESTS 10. International Maritime Organization, ‘Guidelines for

Voluntary Use of the Ship Energy Efficiency Operational
The author declare that they have no known competing Indicator (EEOI)’, 2009.
financial interests or personal relationships that could have
appeared to influence the work reported in this paper. 11. J. Boekhoff, ‘Understand your shipping emissions’ [online,
viewed 08.07.2023] available from https://www.carbonchain.
com/blog/understand-your-shipping-emissions
REFERENCES
12. ‘Ship & Bunker, Rotterdam bunker price’ [online, available
1. Dimerco, ‘IMO 2023 Regulations: Definition and 08.07.2023] available from https://shipandbunker.com/
Implications’ [online], available https://dimerco.com/ prices/emea/nwe/nl-rtm-rotterdam
imo-2023-regulations/
13. ‘Classification Society, Appendix- calculation of energy
2. IMO, ‘Rules on ship carbon intensity and rating system efficiency operational indicator based on operational
enter into force’ [online, viewed 18.06.2023], available from data’ [online viewed 01.07.2023] available from https://

www.imorules.com/GUID-EF0096C9-A807-4C9C-A5EF-
1B0DDE4C01A3.html
14. A. Papanikolaou, ‘Ship Design- Methodology of

Preliminary Design, 2014, Springer Dordrecht, https://
doi.org/10.1007/978-94-017-8751-2;
15. G. Butler, ‘Guide to Ship Repair Estimates’, Oxford, 2000,

Butterworth-Heinemann, Elsevier ISBN 0 7506 4834 1
16. ‘Made in China’ [online] available https://www.made-

in-china.com/products-search/hot-china-products/
Shipbuilding_Plate_Price.html
17. ‘Steel Pipes and fittings, Shipbuilding steel price’

[online viewed 12.07.2023] available from https://www.
sambhavpipes.com/shipbuilding-steel-plate.html

10.2478/pomr-2023-0055
USE OF THE AHP METHOD FOR PREFERENCE

DETERMINATION IN YACHT DESIGN
Jan Sierzputowski *
Artur Karczewski
Przemysław Krata
Gdansk University of Technology, Institute of Naval Architecture, Gdansk, Poland
* Corresponding author: jan.sierzputowski@pg.edu.pl (J. Sierzputowski)
Abstract
A sailing yacht is a human-centred product, the design of which revolves primarily around the wants and desires of
the future owner. In most cases, these preferences are not measurable, such as a personal aesthetic feeling, or a need for
comfort, speed, safety etc. The aims of this paper are to demonstrate that these preferences can be classified and represented
numerically, and to show that they are correlated with the type of yacht owned. As a case study, the owner’s preferences
for deck equipment are considered. These are determined by pairwise comparisons of the importance rankings for features
previously defined by yacht owners, following the analytic hierarchy process (AHP) method. As a result, a quantitative
representation of these preferences is established, and they are shown to be correlated with the type of yacht. The findings
of the current study show that the yacht owners’ preferences can be represented numerically, leading to a utilitarian
conclusion that concerns the support and even some degree of automation of the design process.
Keywords: Yacht Design, AHP Method, Yacht User Preferences, Sailing Yacht, Human-Centred Design
INTRODUCTION safe at sea, economically efficient, and generally perform well

within their scope of application. Although the definitions of
As a technical object, a yacht needs to be designed and built to these requirements are tailored to the purpose, the main thrust
withstand rough sea conditions, to ensure safety and convenience of the designer’s efforts remains very similar [1].
to people on board, and to provide pleasure, or possibly even the The scientific literature in the field of ocean engineering
perception of luxury resulting from spending time in a unique is mainly focused on maritime transport, its impact on the
way. As a recreational vehicle, it needs to meet the expectations environment, and ways to optimise it, with the goals of safety
of users with a wide range of comfort levels. The achievable and health; it therefore includes issues related to improving
speed of a yacht is also a significant feature, since racing forms the efficiency of ships, for example by increasing the speed,
part of the lifestyle of numerous sailors. Whatever the exact capacity, lifespan and human safety while lowering pollution,
purpose, a yacht can be recognised as a human-centred object. costs, risks and accidents [2], [3]. The vast majority of research
This aspect, although clearly more prominent for a yacht than on motor yachts addresses the optimisation of the hull shape
for other watercraft, is to some extent similar for many types of and the automation of the design process [4], [5].
vessels. The main design objectives may vary depending on the For a sailing yacht, the aspects most often studied are its
type of vessel, meaning that the optimal solution depends on behaviour in waves [6] and the prediction of its speed under sail
the specific purpose of the ship. Nevertheless, all ships must be [7]. Other areas of interest are the aero- and hydrodynamics of

the yacht [8], the behaviour of the sails [9], and the performance by the developer of this method. The opinions of shipowners on
of the yacht under different sail settings [10]. Existing articles the use of their yachts, for example sailing them, operating them
have also considered masts and rigging, for example the and maintaining them in good condition, will be considered
overall strength and performance of standing rigging [11], here, with deck equipment and rigging forming the main focus
and the optimisation of the rigging structure design process of the study.
[12]. Although the human factor is considered in several of The ability to determining users’ preferences in a measurable,
the research papers mentioned above, the main purpose of the numerical way could significantly speed up the design process.
researchers has been to improve the yacht as a technical object. The designer could then rely on these numbers to improve their
Yachts are made for people, and designers therefore prefer design. Costs related to the replacement of failed elements, which
a human-centred approach. Human comfort on board is in shipyard conditions can reach up to 40% of the value of the
discussed in scientific publications in terms of ergonomic entire construction [24], would also decrease.
guidelines [13], [14] or with regard to the design of the interior The rest of this paper is organised as follows. In the next
and exterior of a yacht [15]. The term „human factor” is section, the AHP method used in the study is explained, and the
considered mainly in association with safety [16], [17], and rarely study preparation process is presented. The subsequent section
in relation to the conceptual design of the interior of the yacht presents the obtained results and a discussion, while the last
[18]. Although the authors of the latter reference discuss the section presents the conclusion.
importance of knowing the customer’s needs and requirements,
this is not suitable for an automated design process and only
applies to one-off production. The design process itself has been IDENTIFICATION OF THE END-USER’S
considered in several articles, ranging from different design PREFERENCE PATTERNS
approaches and ways of communicating with the customer
[19], through design optimisation [20], to the entire approach As the central idea behind this research is to identify and
to yacht design [21] which involves gathering information quantify the typical patterns of preferences of yacht end-users,
on the client’s preferences in the initial phase of the project. the AHP method was adopted, since there are many examples of
Baranowski in [22], focuses on modifying a sailing yacht to its use in solving a variety of problems where it is impossible to
accommodate disabled individuals. This process entails gaining quantify the decision variables. This method was deemed suitable
a deep understanding of the requirements of such individuals to determine the subjective ratings provided by owners of yachts.
and implementing tailored solutions on the yacht. However, it
must be noted that this scenario is rare and does not extend to Analytic hierarchy process
the production of yachts on a larger scale. The AHP method is a multi-criteria decision support method
The design process is related to the selection of the best (multiple-criteria decision analysis, MCDA) that was proposed
solution for a given project, although this is usually the “best” by Saaty in the 1970s [25]. Methods of this type are used when
only in certain respects, such as cost or a low risk of failure, and the number of decision variables (i.e. the factors influencing
the choice may be made using multi-criteria decision-making the final decision) exceeds human analytical capabilities; in
(MCDM) methods [23]. However, there is a noticeable lack of other words, when there are too many variables for the human
studies of the selection of deck equipment for a yacht in relation mind to be able to grasp them all at once, especially if there are
to the preferences of yacht owners, in other words studies that contradictory features [26]. The AHP is an effective method
take into account a variety of different aspects of yacht operation, for dealing with complex problems of this type [27]. It helps
such as comfort, performance, cost, durability, and aesthetics. decision makers set priorities between alternatives, sub-criteria
These are subject to the individual feelings of the owner, and and criteria in the decision-making process, and to make the best
due consideration has not yet been paid to this subject. Deck decision in a given context [23], [28]–[30]. The method is used
equipment should also be considered as a link between the sailor to structure the problem, starting from the goal to be achieved
and the sail, which can act in both directions. as a result of making a decision, through the criteria affecting
Questions arise as to whether it is possible to objectively the choice, and ending with the possible options.
examine and classify the preferences of sailing yacht owners and
present them numerically, whether there are any correlations
between yachts and their owner’s preferences, and whether these
correlations could be used in the yacht design process.
To answer these questions, the analytic hierarchy process
(AHP) method, a tool that facilitates decision making, is
applied in this study. The aim of this method is to make the
right decision, rather than to indicate certain preferences or to
find any correlations between these and other features. However,
one of the steps of this method requires experts to determine the
weights of the features affecting the decision-making process.
These weights reflect the preferences of the experts. In the case
considered here, they will be determined by the owners of sailing
Fig. 1. The AHP hierarchy pyramid
yachts through a survey created based on the instructions given

The algorithm for the AHP method consists of several Tab. 1. Features of the deck equipment
consecutive steps: Criteria Description
1. D etermining the decision problem;
2. Developing a decision model using a hierarchical structure; Price of the Preferred value: low price.
accessory
3. Comparing criteria in pairs using a fundamental scale
Resistance, failure-free operation, durability,
of comparison; Durability strength, ease of repair of the equipment.
4. Determining priorities and their interpretation; Preferred value: high durability.
5. Dealing with inconsistent answers; Functionality, ease and convenience of use
(efficiency is a general characteristic that allows
6. Group decision making (aggregation of results); Efficiency one to say that one element is better than another
7. Making a decision; in terms of use). Preferred value: high efficiency.
8. Analysing the effects of the decision; Weight Preferred value: low weight.
This process is widely discussed in the scientific literature in of the accessory
relation to many different decision-making problems, including Colour, shape, attractiveness, matching
Aesthetics the appearance of the yacht, overall visual
those relevant to the maritime industry, and will not be further of the accessory and aesthetic impression of the accessory.
explored in this article. For more information, please see the Preferred value: high level of aesthetics.
sources.
• Decision goal: Selection of the optimal deck equipment for
APPLICATION OF THE AHP METHOD IN THIS a given sailing yacht.
STUDY • Criteria influencing the decision: Based on the author’s
experience of sailing and professional work in the selection
Experts and sale of deck equipment, the criteria in Table 1 were
The purpose of using the AHP method is to make the right identified.
decision, rather than to indicate certain preferences of the experts • Selectable options: Groups of accessories should be identified,
or to find any correlations between these preferences and other such as winches, staysail furlers, masts, standing and running
features. This study focuses on the opinions of yacht owners on rigging, cleats, jammers, etc. Individual products should then
the use of their yachts, i.e. sailing them, operating them, and be associated with these groups, and solutions with different
maintaining them in good condition. The experts in this study parameters should be found (e.g. Winch 1, Winch 2, Winch 3,
are therefore all owners of sailing yachts who are responsible etc.). Since the purpose of this study was to examine the
for their maintenance and are their main end-users. preferences of users rather than to select equipment, this step
was omitted.
Decision-making goal The overall hierarchy is shown in the form of a diagram in
The decision-making goal should be formulated in such a way Fig. 2.
that the experts’ answers reflect their personal preferences in In this case, there is one goal involving five equipment features
relation to the use of their yachts. If a decision were to be made, (N = 5) that influence decision making, and 15 options (M = 15)
it should affect the quality of operation of a yacht from the point with different parameters for each feature that classify them
of view of the user (yacht owner). Since the direct link between higher or lower. The options are not considered here.
the sailor and the yacht is the deck equipment and rigging, these
should be the focus. Thus, the decision-making goal in this case Obtaining experts’ judgements
is the selection of the optimal deck equipment for a particular To collect the individual judgements from the owners of
sailing yacht. a wide range of sailing yachts, a survey was created and posted
Hierarchical structure of a decision problem in several social media groups that included yacht owners.
Fig. 2 Hierarchy structure tree of sailing yacht deck equipment applying the AHP method

It was also sent directly to a few yacht owners known to the DETERMINATION OF YACHT
authors. OWNERS’ PRIORITIES
The survey asked experts to compare the features listed in
Table 1, in pairs. As there were five features, 10 comparisons were To establish a priority ranking, each method mentioned in
needed to cover all possible combinations. The respondents were reference [26] was studied, and the outcome with the lowest
asked to decide which of the two features was more important consistency ratio (CR) value [25] was selected to ensure that
than the other, and to what extent, based on a rating scale from the answers did not contradict each other. These priorities
one to nine [25]. The scale was limited to odd numbers only were determined for each respondent. Due to the number of
(1, 3, 5, 7, 9) to make the comparison easier for the respondents. responses involved, only a selection of results are shown in
Furthermore, to check whether the preferences formed some Table 3.
kind of pattern, respondents were asked to provide additional
Tab. 3. Calculated individual priorities
information about the yacht they owned and which formed
the subject of the pairwise comparison, such as the brand, No Price Mass Efficiency Durability Aesthetics CR
hull type, main dimensions and purpose of the yacht. They 1 33.62% 15.19% 17.69% 26.07% 7.43% 0.1697
were also asked to provide brief information on their sailing 2 18.32% 4.76% 13.41% 52.95% 10.56% 0.1605
aspirations and age. 3 3.61% 14.29% 22.50% 50.89% 8.71% 0.2698
...
11 16.40% 19.47% 19.47% 25.19% 19.47% 0.0339
RESULTS
12 26.91% 23.20% 8.61% 29.47% 11.80% 0.4303
Experts’ answers ...
A total of 48 responses were obtained from the experts,
the vast majority of whom were owners of tourist/cruising Verification
yachts. The numbers of respondents for each type of yacht are According to Saaty [25], a CR of greater than 0.1 indicates
presented in Table 2 below. The judgements are shown in the inconsistency, and the answer should be rejected. Unfortunately,
table in Appendix 1. an overwhelming majority of the answers to our survey were
inconsistent (CR > 0.1, highlighted in red in Table 3). To
Tab. 2. Number of respondents by type of yacht owned
investigate these discrepancies, a control participant with
Number the highest CR was consulted, who reported that the priority
Yacht type owned by the respondent of respondents
weights determined by the AHP method on the basis of his
Racing/cruising yachts 2
answers were as he expected; in other words, he considered them
Racing yachts 2 subjectively correct. He also did not identify any substantive
Expedition yachts 5
errors in the construction of the questionnaire.
Seagoing cruising yachts 24 Correcting respondents’ answers

Inland/coastal cruising yachts 15 As the questionnaire was found to be intelligible and
the priority ranking of the results was considered correct,
the WAM method [31] was implemented to improve the CR.
Fig. 3 Yacht owners’ preferences, determined based on aggregated consistent individual judgements by yacht type

Tab. 4. Preferences of sailing yacht owners
Yacht type Price Mass Efficiency Durability Aesthetics CR
Racing/cruising yacht 12.45% 19.15% 33.23% 30.43% 4.75% 0.049
Racing yacht 12.41% 16.45% 33.90% 33.45% 3.78% 0.036
Expedition yacht 16.79% 8.03% 20.75% 42.22% 12.21% 0.026
Seagoing cruising yacht 17.46% 13.87% 17.18% 40.81% 10.68% 0.011
Inland/coastal cruising yacht 19.92% 10.90% 25.85% 27.71% 15.62% 0.005
Tab. 5. Preferences of sailing yacht owners aggregated before CR improvement

Yacht type Price Mass Efficiency Durability Aesthetics CR
Racing/cruising yacht 11.93% 17.76% 33.11% 33.03% 4.16% 0.084
Racing yacht 17.60% 15.56% 31.41% 32.24% 3.20% 0.314
Expedition yacht 16.85% 7.80% 20.31% 43.69% 11.35% 0.061
Seagoing cruising yacht 17.39% 13.53% 16.31% 42.34% 10.43% 0.045
Inland/coastal cruising yacht 19.79% 11.01% 25.01% 28.65% 15.53% 0.022
The individual judgements were corrected, and new priorities impact on the final CR, which was greater than 0.10 only for the
and CR values were calculated. The new results are shown in group of racing yachts. The results obtained from this process
the table in Appendix 2. were very similar to those of the improved answers, as shown
in Fig. 4 and Table 5.
Group aggregation results
After aggregating the results according to the values specific
to each yacht (such as the overall length (LOA), breadth (B), DISCUSSION
displacement (D), etc.), according to the factors of slenderness,
comfort, etc., it was concluded that the purpose of the yacht The use of a nine-point rating scale and a pairwise comparison
had the greatest correlation with the priorities. The individual of features seems to be a good method for determining priorities
judgements were divided into groups related to the type of among yacht owners. The results showed that durability was the
yacht owned, and then aggregated using the geometric mean most important feature for expedition and seagoing cruising
method [32]. As a result, the percentage degree of importance yachts, whereas the efficiency and mass of an accessory were the
was obtained for each of the features of the deck equipment, most important aspects for racing and racing/cruising yachts.
depending on the type of yacht. The results are shown in Fig. 3 The owners of inland/coastal cruising yachts had the most
and Table 4. balanced priorities (regardless of the mass of an accessory) and
In order to explore the impact of improving the respondents’ valued aesthetics most highly.
answers on the relevance of the results, inconsistent judgments The use of the AHP method imposes certain limitations
were also aggregated for comparison. It was found that the on the results. The method used to conduct the study, the
aggregation method using the geometric mean had a positive number of respondents, their questionable proficiency in the
Fig. 4 Yacht owners’ preferences determined based on aggregated inconsistent individual judgements by yacht type before CR improvement

field under study and the consistency of the matrix of their turn, leads to a more accurate determination of the required
answers differ significantly from the recommendations made by nautical qualities, yacht size, equipment etc. When these
the developer of the AHP, which may cast doubt on the results preferences are expressed as numerical values, they can be
obtained in this work. However, as indicated by the control used to automate the selection of certain elements, which can
respondents, the prioritisation of their responses was in line significantly improve the design process.
with their expectations, even if the CR was greatly inflated. Since a relatively small group of yacht owners was involved
Furthermore, a comparison showed that the improvements in this study, it would be advisable to carry out further work
to the CR had a negligible effect on the final results in terms with larger groups. Data collected from a much larger number
of the priorities. The biggest error was found in the priority of of respondents might allow their preferences to be classified
the price reported by racing yacht owners, which reached 5%, according to other factors, such as the sailing area or the size
whereas the error in the efficiency was 2.5% and the variability of the yacht. The use of these data in the design process is
in all the other priorities was less than 1%. This error was due therefore the next step for future work. The classification of
to the small number of respondents in this group (two people) equipment components according to the features presented
and the fact that one of them gave very inconsistent answers. here, and assigning them to particular types of yacht, would
Most of the inconsistent answers that required correction be possible avenues to explore.
came from the owners of tourist yachts. Some of them pointed
to the extreme advantage of feature 1 over feature 2, feature 2
over feature 3, and feature 3 over feature 1. As if all features REFERENCES
were equally important to them. It can therefore be concluded
that people engaged in hobby, tourist and recreational sailing do 1. G. Di Bucchianico and A. Vallicelli, “User-centered approach
not have extensive expert knowledge of the field of equipment for sailing yacht design,” in Human Factors and Ergonomics in
selection, and are not aware of the frequent need to compromise. Consumer Product Design, vol. 2, W. Karwowski, M. Soares,
Much less extreme responses were received from the owners and N. Stanton, Eds., 2011, pp. 445–463.
of racing and expedition yachts, although only two responses
were received from racing yacht owners, one of which was 2. S. Tavakoli, D. Khojasteh, M. Haghani, and S. Hirdaris, “A review
very inconsistent. on the progress and research directions of ocean engineering,”
Research has shown that the preferences of yacht owners in Ocean Engineering, vol. 272, p. 113617, Mar. 2023, doi: 10.1016/j.
terms of equipment vary depending on the purpose of the yacht. oceaneng.2023.113617.
The purpose also affects their sailing aspirations; for example,
an owner may take advantage of the qualities of an expedition 3. T. Song, T. Tan, and G. Han, „Research on preventive
yacht to go on long voyages. This, in turn, affects the owner’s maintenance strategies and systems for in-service ship
experience and expert knowledge. equipment,” Polish Maritime Research, vol. 29, no. 1, pp. 85–96,
The results obtained here indicate that the AHP method Mar. 2022, doi: 10.2478/pomr-2022-0009.
can be used as a priority setting tool. However, the question
to be answered differs from the one that will be posed in the 4. A. Mancuso, A. Saporito, and D. Tumino, “Parametric
method as a decision problem, meaning that it needs to be very hull design with rational Bézier curves,” in Lecture Notes
carefully thought out. in Mechanical Engineering, Springer Science and Business
Media Deutschland GmbH, 2021, pp. 221–227 doi:
10.1007/978-3-030-70566-4_36.
CONCLUSION
5. S. Khan, E. Gunpinar, K. M. Dogan, B. Sener, and P. Kaklis,
The aims of this study were to examine, classify and “ModiYacht: Intelligent CAD tool for parametric, generative,
numerically rank the preferences of sailing yacht owners, and to attributive and interactive modelling of yacht hull forms,” in
explore whether these preferences depended on the type of yacht SNAME 14th International Marine Design Conference, IMDC
owned. The AHP method was applied to the research problem, 2022, Society of Naval Architects and Marine Engineers, 2022,
and it was shown that these preferences were correlated with the doi: 10.5957/IMDC-2022-311.
purpose of the yacht owned by the respondent. A scheme for the
numerical representation of these intangibles was presented, and 6. J. R. Binns and P. A. Brandner, “Dynamic interaction of
may be valuable in the yacht design process, as these data are breaking waves and inverted sailing yachts: Explaining the
much more comprehensible and can be processed by a computer efficacy of mast height retention relative to vertical centre of
in this form. gravity,” Ocean Engineering, vol. 35, no. 17–18, pp. 1759–1768,
The research results show that a group of people who own Dec. 2008, doi: 10.1016/j.oceaneng.2008.08.013.
similar vessels have similar preferences and expectations
towards their yachts and the related equipment. It can therefore 7. G. Guelfi and E. Canepa, “New development in 6-DOF
be concluded that when designing a new yacht, properly algorithms for sailing yacht velocity prediction program and
determining the preferences of the future owner will lead to new insight in appendages force modelling,” Master’s Degree
a more precise determination of the yacht’s purpose. This, in in Nautical Engineering, 2013.

8. C. A. Marchaj, Aero-Hydrodynamics of Sailing. Adlard Coles 22. B. Branowski, M. Zabłocki, P. Kurczewski, and A. Walczak,
Nautical, 2000. „Selected issues in universal design of yachts for people with
disabilities,” Polish Maritime Research, vol. 28, no. 3, pp. 4–15,
9. B. Augier, P. Bot, F. Hauville, and M. Durand, “Dynamic Sep. 2021, doi: 10.2478/pomr-2021-0030.
behaviour of a flexible yacht sail plan,” Ocean Engineering,
vol. 66, pp. 32–43, 2013, doi: 10.1016/j.oceaneng.2013.03.017. 23. E. Miszewska, M. Niedostatkiewicz, and R. Wisniewski,
“The selection of anchoring system for floating houses by
10. T. Matulja, L. Jedretić, and M. Hadjina, “Influence analysis means of AHP method,” Buildings, vol. 10, no. 4, Apr. 2020,
of deck equipment positioning on performances in sailing,” doi: 10.3390/BUILDINGS10040075.
pp. 101–109.
24. M. Naujok, Boat Interior Construction: A Bestselling Guide
11. C. M. Rizzo and D. Boote, “Scantling of mast and rigging of to DIY Interior Boatbuilding, 2nd ed., vol. 1. Adlard Coles
sail boats: A few hints from a test case to develop improved Nautical, 2002.
design procedures,” 2010.
25. T. L. Saaty, The Analytic Hierarchy Process: Planning, Priority
12. M. Pawłusik, R. Szłapczyński, and A. Karczewski, “Optimising Setting, Resource Allocation (Decision Making Series), vol. 1.
rig design for sailing yachts with evolutionary multi-objective McGraw-Hill, 1980.
algorithm,” Polish Maritime Research, vol. 27, no. 4, pp. 36–49,
Dec. 2020, doi: 10.2478/pomr-2020-0064. 26. W. C. H. Beck, A. Prusak, and P. Stefanów, “Budowa i analiza
modeli decyzyjnych krok po kroku AHP-analityczny proces
13. W. Karwowski, M. M. Soares, and N. A. Stanton, “Human hierarchiczny,” 2014.
factors and ergonomics in consumer product design uses
and applications,” CRC Press. 2011. 27. G. Khatwani and A. K. Kar, “Improving the cosine consistency
index for the analytic hierarchy process for solving multi-
14. N. Kalkan, “Human factors and ergonomic considerations criteria decision making problems,” Applied Computing
for super-fast boat design,” 2015. [Online]. Available: http:// and Informatics, vol. 13, no. 2, pp. 118–129, Jul. 2017, doi:
mybroadband.co.za/photos/showfull. 10.1016/j.aci.2016.05.001.
15. S. Ö. Felek, “Parametric sailing yacht exterior and interior 28. A. Jozaghi et al., “A comparative study of the AHP and
design,” 2020, doi: 10.14744/tasarimkuram.2019.30085. TOPSIS techniques for dam site selection using GIS: A case
study of Sistan and Baluchestan Province, Iran,” Geosciences
16. D. Harris, S. Mccartan, B. Verheijden, H. Groningen, and (Switzerland), vol. 8, no. 12, Dec. 2018, doi: 10.3390/
M. Lundh, “European Boat Design Innovation Group: geosciences8120494.
The marine design manifesto. Human factors in safety
management systems view project next generation civil flight 29. F. de Felice and A. Petrillo, “Absolute measurement with
deck concepts view project,” 2014. [Online]. Available: https:// analytic hierarchy process: A case study for Italian racecourse,”
www.researchgate.net/publication/280065046. International Journal of Applied Decision Sciences, vol. 6, no. 3,
p. 209, 2013, doi: 10.1504/IJADS.2013.054931.
17. M. Bilski, “Selected Human Factors in Marina Design,”
Procedia Manuf., vol. 3, pp. 1646–1653, 2015, doi: 10.1016/j. 30. F. de Felice and A. Petrillo, “Multicriteria approach for process
promfg.2015.07.482. modelling in strategic environmental management planning,”
International Journal of Simulation and Process Modelling,
18. T. Bosma, “A human factor’s approach to mega yacht concept vol. 8, no. 1, p. 6, 2013, doi: 10.1504/IJSPM.2013.055190.
design,” Master of Philosophy, University of Strathclyde,
Glasgow, 2013. 31. J. Benítez, X. Delgado-Galván, J. Izquierdo, and R. Pérez-García,
“Improving consistency in AHP decision-making processes,”
19. R. Gill, “Approaches to design,” Des. Stud., vol. 1, no. 3, Appl. Math. Comput., vol. 219, no. 5, pp. 2432–2441, Nov. 2012,
pp. 141–145, Jan. 1980, doi: 10.1016/0142-694X(80)90020-4. doi: 10.1016/j.amc.2012.08.079.
20. P. von Buelow, ‘Using Evolutionary algorithms to aid 32. J. Krejčí and J. Stoklasa, “Aggregation in the analytic hierarchy
designer of architectural structures,” in Creative Evolutionary process: Why weighted geometric mean should be used instead
Systems, Elsevier, 2002, pp. 315–336. doi: 10.1016/ of weighted arithmetic mean,” Expert Syst. Appl., vol. 114,
B978-155860673-9/50050-1. pp. 97–106, Dec. 2018, doi: 10.1016/j.eswa.2018.06.060.
21. L. Larson, R. E. Eliasson, and M. Orych, Podstawy Projektowania

Jachtów, 4th ed., vol. 1. Warszawa: Almapress, 2014.

10.2478/pomr-2023-0056
OPTIMAL UV QUANTITY FOR A BALLAST WATER TREATMENT

SYSTEM FOR COMPLIANCE WITH IMO STANDARDS
Nguyen Dinh Thach

Phan Van Hung *
Faculty of Navigation, Vietnam Maritime University, Haiphong, Viet Nam
* Corresponding author: phanvanhung@vimaru.edu.vn (Phan Van Hung)
Abstract
Ballast water management is an effective measure to ensure that organisms, bacteria and viruses do not migrate
with the ballast water to other areas. In 2004, the International Maritime Organization adopted the International
Convention on the Control and Management of Ballast Water and Ship Sediments, which regulates issues related to
ballast water management. Many technologies have been researched and developed, and of these, the use of UV rays
in combination with filter membranes has been shown to have many advantages and to meet the requirements of
the Convention. However, the use of UV furnaces in ballast water treatment systems requires a very large capacity,
involving the use of many high-power UV lamps. This not only consumes large amounts of electrical energy, but is
also expensive. It is therefore necessary to find an optimal algorithm to enable the UV radiation for the UV controller
in the ballast water sterilisation process to be controlled in a reasonable and effective manner. This controller helps
to prolong the life of the UV lamp, reduce power consumption and ensure effective sterilisation. This paper presents
a UV control algorithm and a controller for a UV furnace for a ballast water treatment system installed on a ship.
The results of tests on vessels illustrate the effect of the proposed UV controller.
Keywords: UV Quantity Controller, Ship, Ballast Water Management, Viable Organisms Threat, Marine Environment.
Nomenclature
UVDose(k) – is the amount of UV at the current point UVbdose=UVdose

j
(k−1) – is the amount of UV in volume
UVDose(k−1) – is the amount of UV at the previous Vj at time k−1
point T – is the sampling time for the
τ – is the amount of water entering and controller
leaving this region dUV – is the difference between two
Va – i s the amount of water injected in one values dUV1 and dUV2
cycle τ Uʹʹ(k) – is the value at the present time
UVadose – is the amount of UV available in the Uʹʹ(k−1) – is the value at the time of the
amount of water injected into the Vj area previous product
Vb = Vj – Va – is the amount of water remaining in
volume Vj at time k−1

INTRODUCTION management systems (G8) were adopted, and were later
transformed into a draft mandatory code for the approval of
Ships use ballast water to ensure stability and ballast water management systems (BWMSs), with particular
manoeuverability. This water is taken in and discharged as reference to the procedure for approval of BWMSs that use
needed to counterbalance the hull stress caused by rough sea active substances (G9).
conditions, loading and unloading operations, or changes in fuel All ships are required to comply with the D2 standard
and water levels. In addition, ballast water helps to control the by 8th September 2024, and must ensure that their BWMSs
trim of the vessel, thereby ensuring it maintains the appropriate meet the required criteria to protect marine ecosystems from
balance and posture during its voyage [1]–[5]. However, this potential invasive species and pathogens [11], [30]. Some
water may harbour a diverse range of organisms, such as essential water quality parameters necessary for effective marine
phytoplankton, zooplankton, bacteria, viruses, and macro- environment management include physico-chemical factors
fauna. The unintentional transfer of potentially invasive alien such as temperature, colour, turbidity, salinity, dissolved oxygen,
species (of which there are around 7,000 to 10,000 different conductivity, suspended solids, and radioactivity. Monitoring
types, including marine microbes, plants, and animals) occurs and understanding these parameters are crucial to ensure the
worldwide on a daily basis, leading to economic losses of tens proper and sustainable management of the marine ecosystem
of billions of US dollars annually [6]–[10]. The International [16], [31]–[34].
Maritime Organization (IMO) [11] has identified numerous Several ballast water treatment systems or combinations of
undesirable and invasive species associated with ballast water systems have been developed and put into practical use to ensure
operations. The introduction of non-indigenous species through compliance with the standards set by the BWM Convention.
ballast water discharge can inflict severe consequences on local It is worth mentioning that ballast water treatment systems that
ecosystems, as invasive species have the potential to outcompete utilise UV rays in conjunction with a membrane filter are highly
native species for resources, leading to significant alterations in regarded for their efficient microbial treatment capability and
the structure and function of an ecosystem [12]–[22]. Moreover, cost-effectiveness [35]–[39].
they have the capacity to introduce diseases and parasites, which Ballast water treatment systems using UV reactors often
can pose a threat to native species and disrupt the delicate employ multiple high-power UV lamps within a single reactor.
balance of the food web. Certain invasive species can even To ensure effective bactericidal results, precise control over the
modify water chemistry, potentially leading to eutrophication UV reactor is necessary, which involves maintaining a consistent
and the proliferation of harmful algal blooms. The impacts of UV lamp dose that adheres to the specified standard. In practice,
ballast water discharge on aquatic ecosystems can also have the dose of UV radiation relies on two critical factors: the flow
significant economic repercussions. The introduction of invasive rate of water passing through the reactor, and the intensity of UV
species can inflict damage on both commercial and recreational radiation within the UV reactor itself [29],[36], [37], [39],[40].
fisheries, diminishing their yields and affecting the livelihoods of Identifying these factors and devising a control method for the
those who depend on them. Furthermore, invasive species can UV reactor is a crucial area of research that requires careful study
cause harm to infrastructure and property, resulting in increased and implementation. At present, the ballast water treatment
maintenance costs and decreased property values. The expenses systems of many global brands employ a basic ON/OFF control
associated with controlling and eradicating invasive species can measure for the UV reactor, although some continuously run
be substantial, and their effects may persist for many years [6], the reactor at full capacity regardless of variations in water flow.
[7], [19],[13], [23]–[29]. As a result, the UV dose becomes unstable, leading to inefficient
The Ballast Water Management Convention (Ballast Water energy consumption and a shorter lifespan for the UV lamps.
Management Convention, 2004) was adopted by an IMO This article focuses on formulating and establishing optimal
Diplomatic Conference in February 2004, and finally came into control equations for the UV reactor, with the aim of addressing
force globally on 8 September 2017. This convention requires the aforementioned drawbacks and improve the performance
ships to effectively treat their ballast water to remove or neutralise of systems such as these [37],[38],[29],[41]-[43].
aquatic organisms and pathogens before discharging it into new
locations. Its aim is to prevent the spread of invasive species
and potentially harmful pathogens. Ships operating under this CONTROL ALGORITHM FOR A UV
convention may be subjected to port state control in any port or FURNACE CONTROLLER
offshore terminal of a party to the Ballast Water Management
Convention; this inspection process may involve verifying the
presence of a valid certificate and an approved ballast water UV QUANTITY MODEL
management plan on board, checking the ballast water record
book, and possibly conducting ballast water sampling in The ballast water of the vessel has zero UV when it enters the
accordance with the guidelines for ballast water sampling (G2) UV furnace. During the process of flowing through the furnace,
to meet standards D1 and D2. the ballast water is treated with UV rays, so the amount of UV
Ballast water management systems (D3) must be approved gradually increases and until the desired amount is reached
by the administration, and must adhere to the IMO guidelines. at the outlet. The distribution of the amount of UV over the
In 2016, revised guidelines for the approval of ballast water length of the furnace is illustrated in Figure 1. The lower the

flow rate of the water, the higher the UV level at the outlet, or For a section Vj of the UV furnace (Figure 4), the amount of
the steeper the characteristic of the UV quantity according to water entering and leaving this region in each calculation cycle τ
the furnace location. is Va = F·τ (litres). In one cycle τ, the amount of UV received
by the ultraviolet lamp is i·τ. The amount of UV in section Vj
at time k (UV dose
j
(k)) can then be approximated as follows:
j b
UV dose
j
(k)= UV dose · Va V+ UV dose · Vb + i · τ (1)
j
Fig. 1. Distributions of UV in the furnace when treating ballast water
To calculate the amount of UV involved, we divide the

UV furnace into equal parts V1, V2, …, Vn. If these are small
enough, the amount of UV at all points in each one of these Fig. 4. Calculation of the amount of UV in a section Vj of the furnace
parts can be considered constant. That is, the amount of UV in
the hypothetical furnace is approximately distributed in each Transforming Eq. (1) gives:
volume fraction Vj as shown in Figure 2. a j
UV dose · F ·τ + UV dose(k–1) · (Vj–F ·τ)
UV dose
j
(k)= Vj +i·τ
(2)
UV dose
j
(k) = UV dose
j
(k–1) · (1 – FV· τ ) +
j
UV dose
a
· FV· τ + i · τ (3)
j
The amount of water pumped out of part Vj of the furnace in

this time period τ will have a UV content equal to the amount
j
of UV in part Vj at the previous time UV dose (k–1).
Fig. 2. Distribution of UV in the furnace according to the hypothesis To ensure the correctness of Eq. (3), the calculation time τ
and the volume of each part Vj must be chosen so that each
A PLC is a computing device based on a microcontroller, time the PLC is updated, the amount of water flowing into each
and the signals from the PLC are digital ones. Calculations with volume Vj does not exceed this volume. In other words, the
application operations for real-time control are performed by condition Fmax.τ < Vj ensures the accuracy of the UV calculation
the PLC over a certain period of time called the calculation in Eq. (3).
cycle τ. If the current time period of the PLC is the kth cycle For systems using PLCs, the calculation period is generally
(Figure 3), then the previous computation times are k−1, k−2, taken as 0.1 s. With a design rated flow rate of 55 litres per
etc. and the subsequent computation times are k+1, k+2, etc. second, the maximum design flow for the furnace is 70 litres
per second. Thus, the volume of each division Vj > 70 · 0.1 = 7
litres. Thus, a UV oven with volume 78 litres can be divided
into 10 parts at most. The smaller the component parts, the
higher the accuracy, but the larger the computational volume;
thus, we need to choose the lowest feasible volume with an
acceptable calculation error.
In this study, we carry out a simulation where the amount
of UV is calculated according to Eq. (3) for a number of
divisions ranging from three to 10. Since the construction of
the model is the same for each number of divisions, we only
Fig. 3. Calculation periods for the PLC
present an illustration for a model with five parts. Using Eq. (3),
As the PLC performs calculations with a period τ, we consider a mathematical model of the amount of UV at time k can be
the variation in the amount of UV between two consecutive constructed as shown in Figure 5, where the UV furnace is
time points. If UVDose(k) is the amount of UV at the current divided into five equal parts, V1 to V5.
point, then UVDose(k−1) is the amount at the previous point.

V5 contains the amount of water coming out of the furnace
after treatment. The variables of the fuzzy control structure are
defined as shown in Figure 7.
Fig. 7. Variables used by the fuzzy controller structure for the UV furnace
The two values eʹ and dUV are determined as follows:

Fig. 5. Example of a model constructed to calculate the amount of UV
in the furnace 1 if E · K1 > 1
eʹ = E · K1 if –1 ≤ E · K1 ≤ 1
To calculate the amount of UV at time k, we need to know –1 if E · K1 < 1
the amount of UV at time k−1. The model in Figure 6 uses (4)
a memory block to save the value at the previous calculation
time. In other words, if the input to the memory block is the 1 if ΔUdose · K2/T > 1
value of UVdose(k), the output is the value of UVdose(k−1). eʹ = ΔUdose· K2/T if –1 ≤ ΔUdose · K2/T ≤ 1
–1 if ΔUdose · K2/T < –1
(5)
The output signal of the fuzzy controller Uʹ for varying

values of the two inputs eʹ and dUV can be calculated in many
ways, for example based on the shape of the membership
function, based on a transformation matrix, or by using an
output signal table.
For PLC devices, constructing membership functions for
the input and output variables requires considerable system
resources and long computation times, which can affect the
controllability of the PLC; hence, to calculate the value of the
fuzzy control output Uʹ we use a table lookup method, as follows.
Fig. 6. Details of the calculation in block Vj
Step 1: Tabulate the output value Uʹ according to the input eʹand

CONSTRUCTING THE CONTROL ALGORITHM dUV
From the fuzzy controller results obtained from the simulation
The entire UV furnace space is divided into five parts, V1, using Matlab software, we enter some representative values in
V2, …, V5, and the amount of UV in each part is defined in the ranges of the input variables eʹ and dUV. The simulation
Eq. (3). V1 receives ballast water from outside the furnace, while then gives the output values Uʹ shown in Table 1.
Tab. 1. Output value Uʹ for varying values of input eʹ and dUV

dUV
Uʹ
–1 –0.8 –0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 1
1 0.5 0.5 0.5 0.5 0.5 0.5 0.3 0.15 0.05 0.02 0
0.8 0.5 0.5 0.5 0.5 0.46 0.38 0.22 0.1 0.03 0 –0.4
0.6 0.5 0.5 0.5 0.5 0.4 0.2 0.1 0.03 0 –0.1 –0.1
0.4 0.5 0.5 0.5 0.36 0.23 0.1 0.03 0 –0.1 –0.4 –0.6
0.2 0.5 0.44 0.36 0.2 0.1 0.03 0 –0.1 –0.3 –0.7 –0.9
eʹ 0 0.5 0.38 0.2 0.1 0.03 0 –0.1 –0.3 –0.8 –0.9 –1
–0.2 0.3 0.22 0.1 0.04 0 –01 –0.4 –0.7 –0.9 –1 –1
0.4 0.15 0.1 0.03 0 –0.1 –0.3 –0.7 –0.9 –1 –1 –1
–0.6 0.05 0.03 0 –0.1 –0.3 –0.8 –0.9 –1 –1 –1 –1
–0.8 0.02 0 –0.1 –0.4 –0.7 –0.9 –1 –1 –1 –1 –1
1 0 –0.1 –0.1 –0.6 –0.9 –1 –1 –1 –1 –1 –1

Step 2: Determine the value of the fuzzy controller output Uʹ
There are two cases in regard to the values of eʹ and dUV at
the input.
Case 1: The values of eʹ and dUV match those in the table, in
which case looking up the value from the table is simple.
The value of the control output Uʹ is exactly the same
as the value in the table.
Case 2: Th
e values of eʹ and dUV do not coincide with any value
in the table, and their values are between two values in
the table. Suppose eʹ is between two values eʹ1 and eʹ2,
and dUV is between two values dUV1 and dUV2. Then,
as eʹ and dUV each have two values, there are four cells
Fig. 9. Method used to determine the value of the control output
in Table 2 below. Uʹ from the lookup table
Tab. 2. Calculation of the value of Uʹ
We apply Eq. (7) to calculate the value of Uʹ on segment
dUV U1U2 in dUV as follows:
Uʹ
dUV1 dUV2
U1 – U2
eʹ1 A1 B1
Uʹ = dUV – dUV · (dUV – dUV2) + U2 (9)
1 2
eʹ
eʹ2 A2 B2 Here, we consider an example where we determine the value
of Uʹ for two inputs, eʹ = 0.1 and dUV = −0.35. In this case, eʹwill
To construct a formula for the lookup table of Uʹ values, we lie between two values eʹ1 = 0.2 and eʹ2 = 0, and dUV between
first derive a formula to determine the value of yK corresponding two values dUV1 = −0.4, dUV2 = −0.2. We can use the lookup
to the value of xk, knowing that the point K(xK, yK) lies between table to get four values, A1, A2, B1, and B2, as shown in Figure 10.
two points M(x1, y1) and N(x2, y2), as shown in Figure 8. The
equation for the line MN has the form:
y – y2
y1 – y2 = xx1––xx22 <–> y = yx11 –– yx22 ·(x –x2) + y2 (6)
Fig. 8. Determining the value of yK in a segment MN

Fig. 10. Lookup table of output values Uʹ
The value of yk is then determined from the value of xk using
Eq. (7): To determine the output signal U, it is necessary to calculate
the value of the output Uʹʹ. Based on the control structure in
yK = yx11 –– yx22 · (xK –x2) + y2 (7) Figure 7, we have:
The method used to determine the control output value Uʹ Uʹʹ(s) = 1s · (K3 · Uʹ(s) + X(s)) (10)
is illustrated in Figure 9. We apply Eq. (7) to calculate the value
of U1 for segment A1A2 and U2 for segment B1B2 depending on where X(s) = Ks (U(s)−Uʹʹ(s)) is the value of the anti-saturation
the value of eʹ, as follows: integral. We convert Eq. (10) to the Z domain using the Tustin
method:
2
U1 = A 1 – A2
eʹ1 – eʹ2 · (eʹ1 – eʹ2) + A2 s = T · zz –+11 (11)
U2 = Beʹ1 ––Beʹ2 · (eʹ1 – eʹ2) + B2 to get Eq. (12) in the Z domain:

1 2
(8)
Uʹʹ(z) = T2 · zz +– 11 · (K3Uʹ(z) + X(z))

DEVELOPMENT OF THE UV QUANTITY
(1– z1 ) · Uʹʹ(z) = T2 · (1+ z1 ) · (K · Uʹ(z) + X(z))
3 (12) CONTROLLER
To implement Eq. (12) in the PLC, we transform it into The basic components of a fuzzy controller are the fuzzy
a differential equation as follows. stage, the composition rule, and the defuzzification stage.
Since basic fuzzy controllers are only capable of processing
T
Uʹʹ(k) – Uʹʹ(k–1) = current signals, these are called static fuzzy controllers. To
2 · (K3 · Uʹ(k) + K3 · Uʹ(k–1) + X(k) + X(k–1)) extend their application to dynamic control problems, the
necessary kinematics are added to the basic fuzzy controller
T
Uʹʹ(k) = Uʹʹ(k–1) + 2 · to provide it with the derivative or integral value of the signal.
(K3 · Uʹ(k) + K3 · Uʹ(k–1) + X(k) + X(k–1)) When used with these kinematics, the basic fuzzy controller
is called a dynamic fuzzy controller.
T
Uʹʹ(k) = Uʹʹ(k–1) + 2 · For the ballast water treatment UV reactor, this is
(K3 · Uʹ(k) + K3 · Uʹ(k–1) + X(k–1)) + T2 · X(k) a nonlinear object. This nonlinearity is expressed in the
(13) relationship between the flow rate F and the amount of UV,
If we set F = Uʹʹ(k–1)+ T2 · (K3 · Uʹ(k) + K3 · Uʹ(k–1) + X(k–1)), then and between the UV lamp control signal (Udk) and the UV
intensity. The relationship between UVdose and the flow F is
T
Uʹʹ(k) = F + 2 · X(k) (14) described in general terms by the system of equations in (13)
and (14). These systems are difficult to control with PIDs, and
To calculate Uʹʹ(k), we consider three cases: experiments are required to determine the accuracy of the
Case 1: Uʹʹ(k) [0,1] controller. In this case, other controllers are often used (for
Then X(k) = ks · (U(k) − Uʹʹ(k)) = 0. From Eq. (14), we have: example based on fuzzy control, neural control, or adaptive
control) for the system. In the present paper, we build a fuzzy
Uʹʹ(k) = F (15) controller, as this has certain outstanding features: (i) the
Case 2: Uʹʹ(k) >1 control is based on the operator’s experience; (ii) there no
Then X(k) = ks · (1 − Uʹʹ(k)), and from Eq. (14) we get: need for an object model to set up the controller; and (iii) it
can be applied to industrial control devices such as PLCs or
T
Uʹʹ(k) = 2 + k2s · T (F + 2 · ks) (16) microprocessors [43].
Case 3: Uʹʹ(k) < 0

Then X(k) = ks · (0 − Uʹʹ(k)), and from Eq. (14), we get:
Uʹʹ(k) = 2 +2F
ks · T (17)
The controller output after the limiting step is determined

by the following system of equations: Fig. 12. Structure fuzzy controller for UV reactor
1 khi Uʹʹ(k) > 1 The fuzzy control structure for the UV reactor is shown in
U(k) = Uʹʹ(k) khi 0 ≤ Uʹʹ(k) ≤ 1 Figure 12. In this structure, the controller has two inputs and
0 khi Uʹʹ(k) < 0 one output signal. The output signal of the controller is in the
(18) range [0,1], corresponding to a UV intensity of the lamp of
The algorithm for calculating the value of the fuzzy controller between zero and Imax.
in the PLC for the UV furnace is summarised in Figure 11. The two inputs of the fuzzy controller are: wrong order
between the UV amount set and the UV amount of water
treated in the reactor. Rate of variation of UV content of ballast
water.
To facilitate the construction of the model and the signal
processing step, all of the input values to the fuzzy controller
are converted to standard values between −1 and 1. To do
this, we need to pass additional inputs in the form of the
coefficients K1 and K 2 , as shown in Figure 13.
Fig. 11. Algorithm for calculating the output value of the fuzzy control

Fig. 13. Fuzzy control structure with additional inputs in the form
of value conversion factors
We then determine the conversion factor K1 for the control
error input e. The coefficient K1 is used to convert the range of
the deviation e from [−e0, e0] to the range [−1, 1] for input into Fig. 14. Fuzzy controller toolbox with the membership function
of the input bias UV
the fuzzy controller. The range [−e0, e0] should be chosen so that
for the smallest value of the limit e0, the change in e is greatest After building the membership function for the inputs, we
in this interval. constructed the membership function for the outputs. From the
To ensure
dm
the good operation and impact of the bias UV intensity response characteristics of the system, we know that
e0 = UV
10
dose
, the coefficient K1 is determined using Eq. (19): the rate of increase in the UV intensity is limited by the response
of the lamp, meaning that this is a factor that cannot be further
10 10
K1 = e10 = UV dm =
200
= 0.05 (19) increased. The rate of reduction in UV intensity is also high. We
dose
therefore created an asymmetric controller characteristic in which
We determine the conversion factor K2 for the variable rate the control signal used to increase the UV intensity changed more
input UVdose as follows. K2 is used to convert the range of values slowly than the rate of decrease in the UV intensity of the reactor.
for the rate of change dUVdose/dt from [−dUV0, dUV0] to the range This meant that the control characteristic could eliminate the
[−1,1] to feed into the fuzzy controller. Based on the simulated problem whereby the system response cannot keep up with the
response of the rate of change of UV in the reactor, the rate of changing speed of the control, which affects the output quality of
change of UVdose at the fastest time has the value dUVdose/dt = 10. the system. To build the output signal, we used language variables
Thus, the dUVdose/dt variable takes values in the range [−10, 10], for the output with the values shown in Table 3.
or in other words, the coefficient K2 = 1/10 = 0.1. When the two inputs (e, dUVdose/dt) and the fuzzy rule had
The construction of the fuzzy controller was first carried out been built, the controller was able to completely determine the
in Matlab to simulate and verify the operation before applying explicit value of the output Udk according to the values of the
it to real objects. Simulation can help to shorten the equipment input. The relationship between Uʹ and the two input signals
testing time, reduce costs and allow us to gain experience in (e, dUVdose/dt) is shown graphically in Figure 10. In this feature,
controlling the system. Here, we used a Sugeno fuzzy controller Output 1 of the controller varies in the range [−1, 0.5]. With these
[46] with two inputs (the signal bias e and the variable dUVdose/dt) characteristics, the controller will have a slower increase rate of
and one output (Output 1). This controller was implemented 1/2 than the decrease rate of the signal. As mentioned above, this
in Matlab with the Fuzzy Toolbox, as shown in Figure 14. The deviation in the rate of increase/decrease is found purely based
next step was to build linguistic variables for the inputs/outputs on practical experience, and depends on the rates of increase and
and to create membership functions corresponding to these decrease in the UV intensity in the reactor being asymmetric.
linguistic variables for each input/output. To simplify the process This surface is also the basis for building algorithms for fuzzy
of building the membership function for the input linguistic controllers on PLC.
variables, we define linguistic values as shown in Table 3.
Tab. 3. Linguistic variables for signal input
Linguistic variable Meaning Value

GN Great negative –1
LN Large negative –0.6
SN Small negative –0.3
ZE Zero 0
SP Small positive 0.1
LP Large positive 0.3
GP Great positive 0.5
Fig. 15. Developing the fuzzy rule in Matlab software

Tab. 4. Fuzzy rule for controller a designated control tank on the ship. This control tank
dUVdose/dt serves as a reference point for monitoring the quality and
Output 1 condition of the ballast water before it undergoes treatment.
GN LN SN Z SP LP GP
• Pumping ballast water into the ballast tanks for treatment:
GN Z SN LN GN GN GN GN
After filling the control tank, the next step is to pump ballast
LN SP Z SN LN GN GN GN
water from this tank into the ship’s ballast tanks, which
SN LP SP Z SN LN GN GN form part of the treatment system. It is at this point that
e Z GP LP SP Z SN LN GN the ballast water will be subjected to the treatment process
SP GP GP LP SP Z SN LN designed to eliminate or reduce the concentration of aquatic
LP GP GP GP LP SP Z SN organisms.
GP GP GP GP GP LP SP Z
• Draining the ballast water from the control tank: Once
the ballast water has been transferred to the tanks of the
With the proposed control structure, after building the treatment system, the control tank is emptied or drained.
component models, we get the simulation model shown in This step is essential for maintaining consistency in the
Figure 11 and Table 3, where the input is in the range [−1,1], and testing process, as it allows for comparison between treated
the controller output after the integral is in the range [0%,100%]. and untreated water samples.
• Discharging ballast water from treated ballast tanks:
Following the treatment process, the treated ballast water
EVALUATION OF BIOCHEMICAL is discharged from the tanks. In this step, the effectiveness
EFFICIENCY DURING TESTING OF of the treatment system is evaluated in terms of rendering
BALLAST WATER TREATMENT SYSTEM the ballast water compliant with environmental regulations,
particularly concerning the control of invasive species and
ACCORDING TO IMO STANDARD G8 pathogens.
To meet the IMO G8 requirements, it is crucial that this
sequence of work steps is repeated continuously for at least six
CONFIGURATION FOR ON-BOARD TESTING months. The testing process must also yield a minimum of three
consecutive successful test cycles to demonstrate the consistent
The IMO G8 provision pertains to the testing of ballast and reliable performance of the ballast water treatment system
water treatment systems on board vessels. According to these for various operating conditions and environmental factors.
regulations, testing of a ballast water treatment system requires This stringent testing protocol helps ensure that ships’ BWMSs
the installation of such a system on a ship where the normal meet international standards for environmental protection.
ballast operations are carried out. This means that the ballast
system of the ship should be fully operational in the usual BALLAST WATER SAMPLING AND ANALYSIS
manner. DURING TESTING
To conduct tests on the proposed ballast water treatment
system, a specific procedure was followed. The ballast water (1) Number of samples in one test cycle
treatment system was installed on the ship, and was configured Three input water samples are taken during the process of
as shown in Figure 16. This configuration diagram served as pumping ballast water into the counter-egg tank (taken at the
a guide for the proper installation and arrangement of the beginning, middle and end of the process).
system components to ensure that it functioned as intended.
This setup allowed for the systematic testing and evaluation
of the system’s performance under real-world conditions, and
ensured that it met the requirements and standards set out in
the IMO G8 regulations.
ONBOARD TEST PROCEDURE
In accordance with the requirements outlined in G8,

shipboard testing of ballast water treatment systems is
a comprehensive process that is conducted over a minimum
duration of six months. During this testing period, there are
specific criteria that must be met, and a series of test cycles
need to be completed successfully. Each test cycle comprises
several distinct steps, which ensure a thorough evaluation of
the performance of the treatment system, as follows:
• Pumping ballast water into the control tank: The initial
step of the test cycle involves pumping ballast water into Fig. 16. Diagram showing an example of a test configuration for a ballast
water treatment system on a ship

Three samples are taken from the discharge line during standard are presented in Table 5, where the average value of
the process of discharging the ballast water from the egg tank a parameter for three samples is recorded in the corresponding
(taken at the beginning, middle and end stages). A total of nine cell in the results.
samples are taken from the discharge line during the process The test results in Table 5 show that the input parameters
of discharging the ballast water from the treated ballast tank for the water were in accordance with the requirements of the
(three samples in the first stage, three in the middle stage, and IMO. The parameters for the ballast water discharge from the
three in the final stage). reference tank are also in accordance with the requirements of
IMO, and the parameters of the treated water from the ballast
(2) Test criteria for samples tank met the IMO standard D2.
For each sample, the following criteria need to be met: The measurement results corresponding to the actual
Environmental parameters: temperature, salinity (PSU), amounts of UV when operating on board under different
TSS (mg/l), DOC (mg/l) and POC (mg/l); conditions show that the controller works well, and the
Number of living organisms: 50 μm/m3; parameters of the system are suitable at the design value. The
Number of living organisms: 10–50 μm/1 ml; Vietnam Institute of Marine Environment and Resources has
Vibrio cholerae (cholera): cfu/100 ml; confirmed that various samples of ballast water after treatment
Escherichia coli group (intestinal bacilli): cfu/100 ml; from the commissioning testing conducted with the subject
Intestinal enterococci group: cfu/100 ml; BWMS are considered to comply with D2 discharge standard
Heterotrophic bacteria: cfu/1 ml. (IMO BWM Convention).
(3) Criteria for a successful test cycle

Step 1: Inlet water requirement CONCLUSION
The input water must satisfy the following criteria:
Number of living organisms with size ≥ 50μm: 100/m3; Ballast water management is a vital measure to ensure that
Number of living organisms 10–50 μm in size: 100/1 ml organisms, bacteria, and viruses do not spread to new areas
Step 2: Requirements for water samples during discharge of the when discharging ballast, in accordance with the International
ballast water from the control tank Convention on the Control and Management of Ballast Water
Number of living organisms 50 μm in size: 10/m3; and Ship Sediments established in 2004, which regulates ballast
Number of living organisms 10–50μm in size: 10/1 ml water management issues. The UV reactor used in a ballast
Step 3: The requirements for water samples during the discharge water treatment system often requires multiple high-power
of ballast water from the treated ballast tank are UV lamps within a single unit, due to its large capacity. These
considered to comply with the D2 discharge standard high-power UV lamps consume significant electrical energy,
(IMO BWM Convention). and impose high costs.
Efficiently controlling UV radiation during the water
disinfection process is therefore essential to extend the lifespan
EXPERIMENTAL RESULTS of UV lamps, reduce energy consumption, and ensure effective
AND DISCUSSION antimicrobial treatment. This research has focused on developing
optimal equations 19 (and Eqs. (4)–(18)) and a controller for
On-board testing of the ballast water treatment system was a UV reactor in a ballast water treatment system (Figs. 11–13)
completed over a test period exceeding six months. Testing installed on a ship. Experimental results from five ships have
was carried out in accordance with the IMO Guidelines for been presented to demonstrate the efficiency achieved by the
the Approval of Ballast Water Treatment Systems (G8). The proposed UV controller. When the proposed control algorithm
collection and analysis of ballast water samples was carried is applied, the UV lamp shines at the appropriate intensity
out by the Institute of Marine Environment and Resources to meet the ballast water treatment requirements, rather
under the supervision of the Vietnam Register. The test results than continuously operating at high intensity. This results in
for the ballast water treatment system according to IMO’s G8 significant energy savings compared to a system without a UV
Tab. 5. Measured values
IMO standard
Treatment rated capacity
Name of ship Viable organisms ≥50μm Viable organisms 10-50μm Sampling period
(m3/h)
(org/m3): <10 (org/ml): <10
ANBIEN BAY 500 4 6.8 10h50–14h00, 02/06/2022
TTC PIONEER 200 No live organisms detected 7.8 11h10–12h15, 06/06/2022
ROYAL 89 100 6 4.5 09h50–10h40, 15/08/2022
TRACY 150 1 2.9 15h00–15h30, 20/09/2022
AN THINH PHU 08 100 1 0.2 08h00–09h00, 14/10/2022
THAI BINH 35 50 1 1.7 11h30–12h10, 02/11/2022

controller. Furthermore, operating at low intensity rather than 7. P. E. Neill and M. Arim, “Human Health Link to Invasive
continuously at high intensity can help prolong the life of the Species,” in Encyclopedia of Environmental Health, pp. 570–578,
UV lamp. 2019, doi: 10.1016/B978-0-12-409548-9.11731-2.
The development of effective treatment approaches such as
these can enhance ballast water management, thus mitigating the 8. S. Gollasch, C. L. Hewitt, S. Bailey, and M. David, “Introductions
negative impacts of invasive species on marine environments, and transfers of species by ballast water in the Adriatic Sea,”
economies, and human well-being. Mar. Pollut. Bull., vol. 147, pp. 8–15, Oct. 2019, doi: 10.1016/j.
marpolbul.2018.08.054.
Acknowledgments
9. O. Vidjak et al., “Zooplankton in Adriatic port environments:
The authors would like to thank their colleagues at Vietnam Indigenous communities and non-indigenous species,” Mar.
Maritime University, Vietnam Register, Institute of Marine Pollut. Bull., vol. 147, pp. 133–149, Oct. 2019, doi: 10.1016/j.
Environment and Resources, and the shipping companies for marpolbul.2018.06.055.
their assistance in this work. We also appreciate the constructive
suggestions from the anonymous reviewers, which were 10. P. Van Hung, K.-S. Kim, and M. Lee, “Cooperative response
invaluable in improving the quality of the manuscript. to marine hazardous and noxious substances and oil spill
incidents in the ASEAN region,” Australian Journal of
Maritime & Ocean Affairs, vol. 11, no. 1, pp. 61–72, Jan. 2019,
References doi: 10.1080/18366503.2018.1559524.
1. G. Altug, S. Gurun, M. Cardak, P. S. Ciftci, and S. Kalkan, 11. IMO, International Maritime Organization [IMO], 2004.
“The occurrence of pathogenic bacteria in some ships’ ballast International Convention for the Control and Management
water incoming from various marine regions to the Sea of of Ships’ Ballast Water and Sediments. International
Marmara, Turkey,” Mar. Environ. Res., vol. 81, pp. 35–42, Maritime Organization, London. [Online]. Available: http://
Oct. 2012, doi: 10.1016/j.marenvres.2012.08.005. www.imo.org/en/About/Conventions/List ofConventions/
Pages/International- Convention-for-the-Control-and-
2. L. J. Aridgides, M. A. Doblin, T. Berke, F. C. Dobbs, Management-of-Ships%27-Ballast-Water-and-Sediments-
D. O. Matson, and L. A. Drake, “Multiplex PCR allows (BWM).aspx
simultaneous detection of pathogens in ships’ ballast water,”
Mar. Pollut Bull., vol. 48, no. 11, pp. 1096–1101, Jun. 2004, 12. G. Elidolu, S. I. Sezer, E. Akyuz, O. Arslan, and Y. Arslanoglu,
doi: 10.1016/j.marpolbul.2003.12.017. “Operational risk assessment of ballasting and de-ballasting
on-board tanker ship under FMECA extended evidential
3. J. Chen, Y. Lin, J. Zhou Huo, M. Xia Zhang, and Z. Shang Ji, reasoning (ER) and rule-based Bayesian network (RBN)
“Optimization of ship’s subdivision arrangement for offshore approach,” Reliab. Eng. Syst. Saf., vol. 231, p. 108975, Mar. 2023,
sequential ballast water exchange using a non-dominated doi: 10.1016/j.ress.2022.108975.
sorting genetic algorithm,” Ocean Engineering, vol. 37, no. 11,
pp. 978–988, Aug. 2010, doi: 10.1016/j.oceaneng.2010.03.012. 13. D. K. Gray, I. C. Duggan, and H. J. Macisaac, “Can sodium
hypochlorite reduce the risk of species introductions from
4. I. Staffell and P. Balcombe, “How to decarbonise international diapausing invertebrate eggs in non-ballasted ships?,” Mar.
shipping: options for fuels, technologies and policies,” Energy Pollut. Bull., vol. 52, no. 6, pp. 689–695, Jun. 2006, doi:
Convers. Manag., [Online]. Available: https://www.academia. 10.1016/j.marpolbul.2005.11.001.
edu/38669854/How_to_decarbonise_international_
shipping_options_for_fuels_technologies_and_policies 14. P. P. Sangave, A. Mukherjee, and A. Pandit, “Ballast water
treatment using hydrodynamic cavitation,” Nov. 2014.
5. P. Van Hung, K.-S. Kim, L. Q. Tien, and N. M. Cuong, [Online]. Available: https://www.semanticscholar.org/paper/
“Distribution of oil spill response capability through Ballast-Water-Treatment-Using-Hydrodynamic-Sangave-
considering probable incident, environmental sensitivity Mukherjee/ ca3f5edcc2a28ba175b1a4c0093b70b49bb2a334
and geographical weather in Vietnamese waters,” Journal
of International Maritime Safety, Environmental Affairs, 15. S. Rahman, “Implementation of ballast water management
and Shipping, vol. 2, no. 1, pp. 31–41, Nov. 2018, doi: plan in ships through ballast water exchange system,”
10.1080/25725084.2018.1511240. Procedia Eng., vol. 194, pp. 323–329, Jan. 2017, doi: 10.1016/j.
proeng.2017.08.152.
6. G. Drillet et al., “Improvement in compliance of ships’
ballast water discharges during commissioning tests,” Mar. 16. E. Tsolaki and E. Diamadopoulos, “Technologies for ballast
Pollut. Bull., vol. 191, p. 114911, Jun. 2023, doi: 10.1016/j. water treatment: A review,” Journal of Chemical Technology
marpolbul.2023.114911. & Biotechnology, vol. 85, no. 1, pp. 19–32, Jan. 2010, doi:
10.1002/jctb.2276.

17. S. A. Bailey and H. Rajakaruna, “Optimizing methods to 29. J.-T. Baek, J.-H. Hong, M. Tayyab, D.-W. Kim, P. R. Jeon, and
estimate zooplankton concentration based on generalized C.-H. Lee, “Continuous bubble reactor using carbon dioxide
patterns of patchiness inside ballast tanks and ballast water and its mixtures for ballast water treatment,” Water Res.,
discharges,” Ecol. Evol., vol. 7, no. 22, pp. 9689–9698, Nov. vol. 154, pp. 316–326, 2019, doi: https://doi.org/10.1016/j.
2017, doi: 10.1002/ece3.3498. watres.2019.02.014.
18. C. Grob and B. G. Pollet, “Regrowth in ship’s ballast water 30. IMO, “Globallast guidelines for development of a national
tanks: Think again!” Mar. Pollut. Bull., vol. 109, no. 1, ballast water management Strategy, 1997. Guidelines for
pp. 46–48, Aug. 2016, doi: 10.1016/j.marpolbul.2016.04.061. preventing the introduction of unwanted organisms and
pathogens from ships ballast waters and sediment discharges.
19. V. Raţᾰ and L. Rusu, “Ballast water pollution risk assessment in Resolution A.868 (20).” 2004.
the Black Sea,” in Mechanical Testing and Diagnosis, Jan. 2021,
pp. 35–40. doi: 10.35219/mtd.2020.4.05. 31. G. H. Briton, B. Yao, and G. Ado, “Evaluation of the Abidjan
lagoon pollution,” Journal of Applied Sciences and Environmental
20. E. Briski et al., “Combining ballast water exchange and Management, vol. 11, no. 2, 2007, doi: 10.4314/jasem.
treatment to maximize prevention of species introductions to v11i2.55030.
freshwater ecosystems,” Environ. Sci. Technol., vol. 49, no. 16,
pp. 9566–9573, Aug. 2015, doi: 10.1021/acs.est.5b01795. 32. J. Liu, P. Wang, G. Liu, J. Dai, J. Xiao, and H. Liu, “Study of the
characteristics of ballast bed resistance for different temperature
21. A. Travizi et al., “Macrozoobenthos in the Adriatic Sea and humidity conditions,” Constr. Build. Mater., vol. 266,
ports: Soft-bottom communities with an overview of non- p. 121115, Jan. 2021, doi: 10.1016/j.conbuildmat.2020.121115.
indigenous species,” Mar. Pollut. Bull., vol. 147, pp. 159–170,
Oct. 2019, doi: 10.1016/j.marpolbul.2019.01.016. 33. N. A. Salleh et al., “Pathogenic hitchhiker diversity on
international ships’ ballast water at West Malaysia port,” Mar.
22. P. Mozetič et al., “Phytoplankton diversity in Adriatic ports: Pollut. Bull., vol. 172, p. 112850, Nov. 2021, doi: 10.1016/j.
Lessons from the port baseline survey for the management of marpolbul.2021.112850.
harmful algal species,” Mar. Pollut. Bull., vol. 147, pp. 117–132,
Oct. 2019, doi: 10.1016/j.marpolbul.2017.12.029. 34. G. Romanelli et al., “Ballast water management system:
Assessment of chemical quality status of several ports in
23. J.-H. Park, Y.-B. Sim, S.-Y. Kang, and S.-H. Kim, “Inactivation Adriatic Sea,” Mar. Pollut. Bull., vol. 147, pp. 86–97, Oct. 2019,
of indicating microorganisms in ballast water using chlorine doi: 10.1016/j.marpolbul.2017.12.030.
dioxide,” Ecology and Resilient Infrastructure, vol. 5, no. 3,
pp. 111–117, Sep. 2018, doi: 10.17820/ERI.2018.5.3.111. 35. M. R. First and L. A. Drake, “Life after treatment: detecting
living microorganisms following exposure to UV light and
24. T. McCollin, G. Quilez-Badia, K. D. Josefsen, M. E. Gill, chlorine dioxide,” J. Appl. Phycol., vol. 26, no. 1, pp. 227–235,
E. Mesbahi, and C. L. J. Frid, “Ship board testing of Feb. 2014, doi: 10.1007/s10811-013-0049-9.
a deoxygenation ballast water treatment,” Mar. Pollut. Bull.,
vol. 54, no. 8, pp. 1170–1178, Aug. 2007, doi: 10.1016/j. 36. Z. Manxia, L. Shengjie, T. Xiaojia, L. Xiang, and Z. Yimin,
marpolbul.2007.04.013. “Evaluation of micro-pore ceramic filtration and uv radiation
combination on ballast water treatment,” in 2010 International
25. L. A. Drake, M. A. Doblin, and F. C. Dobbs, “Potential Conference on Digital Manufacturing & Automation,
microbial bioinvasions via ships’ ballast water, sediment, pp. 670–673, Dec. 2010, doi: 10.1109/ICDMA.2010.374.
and biofilm,” Mar. Pollut. Bull., vol. 55, no. 7, pp. 333–341,
Jan. 2007, doi: 10.1016/j.marpolbul.2006.11.007. 37. P. P. Stehouwer, A. Buma, and L. Peperzak, “A comparison
of six different ballast water treatment systems based on
26. S. Vodyanitskaya et al., “Methods of decontamination of ship UV radiation, electrochlorination and chlorine dioxide,”
ballast water with polyguanidine disinfectant,” International Environ. Technol., vol. 36, no. 16, pp. 2094–2104, Aug. 2015,
Journal of Infectious Diseases, vol. 79, p. 77, Feb. 2019, doi: doi: 10.1080/09593330.2015.1021858.
10.1016/j.ijid.2018.11.195.
38. I. Rivas-Zaballos, L. Romero-Martínez, I. Moreno-Garrido,
27. E. Lakshmi, M. Priya, and V. S. Achari, “An overview on the J. Moreno-Andrés, A. Acevedo-Merino, and E. Nebot,
treatment of ballast water in ships,” Ocean Coast Manag., vol. 199, “UV-LEDs combined with persulfate salts as a method to
p. 105296, Jan. 2021, doi: 10.1016/j.ocecoaman.2020.105296. inactivate microalgae in ballast water,” Journal of Water Process
Engineering, vol. 51, p. 103361, Feb. 2023, doi: 10.1016/j.
28. M. N. Tamburri, K. Wasson, and M. Matsuda, “Ballast water jwpe.2022.103361.
deoxygenation can prevent aquatic introductions while
reducing ship corrosion,” Biol. Conserv., vol. 103, no. 3, 39. J. Xiao, Y. Xu, L. Hu, and H. Wu, “Evaluating the treatment
pp. 331–341, Mar. 2002, doi: 10.1016/S0006-3207(01)00144-6. performance of filtration & real-time UV irradiation

processes for bacteria and pathogens in fresh ballast water,”
Reg. Stud. Mar. Sci., vol. 63, p. 102971, 2023, doi: https://doi.
org/10.1016/j.rsma.2023.102971.
40. N. F. Gray, “Chapter Thirty-Four - Ultraviolet disinfection,”

in Microbiology of Waterborne Diseases (Second Edition),
S. L. Percival, M. V Yates, D. W. Williams, R. M. Chalmers,
and N. F. Gray, Eds., Second Edition. London: Academic
Press, 2014, pp. 617–630. doi: https://doi.org/10.1016/
B978-0-12-415846-7.00034-2.
41. Y. Wang, L. Zou, L. Ma, Z. Zhao, and J. Guo, “A survey

on control for Takagi-Sugeno fuzzy systems subject to
engineering-oriented complexities,” Systems Science &
Control Engineering, vol. 9, no. 1, pp. 334–349, 2021, doi:
10.1080/21642583.2021.1907259.
42. “DESMI to manufacture ballast water treatment system under

licence,” Pump Industry Analyst, vol. 2012, no. 8, pp. 12–13,
2012, doi: https://doi.org/10.1016/S1359-6128(12)70370-1.
43. B. Sayinli, Y. Dong, Y. Park, A. Bhatnagar, and M. Sillanpää,

“Recent progress and challenges facing ballast water
treatment – A review,” Chemosphere, vol. 291, p. 132776, 2022,
doi: https://doi.org/10.1016/j.chemosphere.2021.132776.

10.2478/pomr-2023-0057
EFFECTS OF SWAY AND ROLL EXCITATIONS ON SLOSHING LOADS

IN A KC-1 MEMBRANE LNG TANK
Se-yeol An1
Hyeon-won Jeong1
Ohyoung Kim2
W. Jaewoo Shim1,*
1
Department of Chemical Engineering, Dankook University, Gyeonggi-do, 16890, South Korea
2
School of Polymer Science and Engineering, Dankook University, Gyeonggi-do, 16890, South Korea
* Corresponding author: wjshim@dankook.ac.kr (W. Jaewoo Shim)
Abstract
This study investigates the effects of sway and roll excitations on sloshing liquid loads in a tank, using Ansys Fluent
software. The model considered in the study is a 1:50 scaled membrane-type tank, based on a KC-1 membrane LNG
tank designed by Korea Gas Corporation (KOGAS). The volume of fluid (VOF) method is used to track the free surface
inside the tank, and the standard k-ε model is applied to express the turbulent flow of the liquid. To explore the motion
of the tank under excitation, a user-defined function (UDF) and a dynamic mesh technique are employed to control
the external forces exerted on the tank through its motion. The results, in the form of time series data on the sloshing
pressures in the tank under pure sway, roll, and coupled sway-roll, are analysed, with specific ranges for the excitation
amplitudes and frequencies. We show that variations in excitation frequency and amplitude significantly influence
the sloshing loads. Sloshing loads are found to intensify when the excitation frequency matches the tank’s primary
natural frequency, 1.0 ω'1. Furthermore, with coupled sway-roll excitations, the sloshing loads are weakened when the
sway and roll are in-phase and are intensified when these are out-of-phase. Fast Fourier transform analysis provides
insights into the frequency domain, showing that the dominant frequency is 0.88 Hz and it is approximately equal to
the tank’s primary natural frequency, 1.0 ω'1.
Keywords: KC-1 membrane LNG tank; Sloshing; Sway; Roll; Coupled sway-roll; Fast Fourier transform (FFT)
introduction tank [1]. Consequently, in severe cases, the loads generated

by sloshing can lead to structural damage to the tank’s
The term ‘sloshing’ refers to unrestrained movements of
0 inner walls. These issues are related to ensuring the safety
the free surface of a liquid in a container, and occurs when of waterway transportation of liquefied natural gas (LNG)
a partially filled tank is exposed to external perturbations, in LNG tankers. To address these concerns, it is important
which cause the free surface of the liquid to oscillate [1]. In to predict the sloshing flow of a liquid loaded in a tank
the case of small perturbations, sloshing can be approximated and the resulting pressure variations. Accurate predictions
as a linear superposition of different wave components [2]. and an understanding of sloshing behaviour can help in
However, it is important to note that sloshing flow represents devising measures to mitigate potential structural damage
a complex fluid motion. Under certain conditions, the and enhance the safety of LNG transportation.
pressure exerted on the inner walls of the tank that are The motion of a tank floating on the sea has six degrees of
in contact with the liquid can increase dramatically; this freedom (6DOF), with external forces exerted on the tank that
is because the unrestrained free surface tends to undergo can include a single DOF, such as surge, sway, heave, pitch,
significant excursions for even small movements of the roll, or yaw as well as multiple coupled DOFs [3]. As a result,

many studies have been conducted to investigate sloshing COMPUTATIONAL MODELING
within tanks subjected to multiple coupled excitations [2–5].
Chen and Wu [3] conducted a study of fluid sloshing in
three-dimensional tanks filled to an arbitrary depth under COMPUTATIONAL MODEL
various excitation frequencies and motion with multiple
DOFs. They demonstrated that instability of sloshing occurs In a sloshing analysis, the free surface of a liquid exhibits
when the excitation frequency of the heave motion is twice nonlinearity as the external forces increase, and one
as large as the fundamental natural frequency under coupled challenging aspect of sloshing analysis is the accurate tracking
surge-sway-heave excitations. Hou et al. [2] studied liquid of the flow of this nonlinear free surface. To address this issue,
sloshing in a two-dimensional tank under single and multiple Hirt and Nichols [7] introduced the volume of fluid (VOF)
coupled external excitations, such as coupled sway-roll and method, which represents the volume occupied by the liquid
coupled sway-roll-heave excitations. In their research, they in each computational cell as a discrete function. This method
showed that sloshing loads are intensified when the tank provides a solution for effectively tracking the nonlinear free
experiences multiple coupled excitations and that match
(VOF) method, which represents surface waves. Since its occupied
the volume introduction, bythetheVOF
liquid method
in eachhas compu
the first order natural frequency of the tank. Wu et al. [4] been widely used to track the free surface in sloshing analyses,
conducted experiments to investigate discrete function.
the impact This method
loads from and it isprovides
also adopted a solution for effectively
in the present tracking
study, using Ansys the nonlin
Fluent
liquid sloshing at low liquid loadingwaves.
ratesSince
under its introduction,
coupled software. theEq.VOF method
(1) gives has been
the function widely the
that expresses used to track the
volume
roll-pitch excitations. They observedsloshing
that whenanalyses,
the excitation
and it fraction
is also of liquid occupying
adopted in the present a cell. study, using Ansys Fluent s
frequencies for roll and pitch are between
gives the 0.98 f1 and 1.113
function that expresses the volume fraction of liquid occupying a cell.
f1, where f1 represents the first-order natural frequency of
a sloshing liquid in a rectangular tank, a roof-bursting 0
phenomenon by the liquid occurred. In addition, they noted 𝛼𝛼(𝑥𝑥, 𝑦𝑦, 𝑧𝑧) = {0 ~ 1 (1)
that the maximum impact pressure was observed at 1.09 f1. 1
In summary, several studies have been conducted on
sloshing in tanks under various liquid When the levels
filling valueandof α is zero, When this
therepresents
value of α is anzero,empty state with
this represents no liquid
an empty statein the c
of α between
ship motions. However, limited attention has beenzerogivenand
to one with no liquidainfree
represent the cell, whilethat
surface values of α between
includes both zero liquidandand gas.
investigating sloshing phenomenaofunder the influence of one represent a free surface that
α is one, this indicates that only liquid is present in the cell. Fig. 1 showsincludes both liquid and gas. the vol
coupled sway-roll excitations, which represent one of the most When the value of α is one, this indicates that only liquid is
liquid in each of the cells based on the values of 𝛼𝛼.
extreme instances of sloshing-induced structural damage present in the cell. Fig. 1 shows the volume fractions of liquid
within the context of the 6DOF motion of a tank [5]. The in each of the cells based on the values of α.
largest sloshing loads occur in the range 0.1 ≤ h/H ≤ 0.5 for
the liquid filling height [5]. Hence, this study investigates
sloshing in a tank filled to 50% with water under pure
sway, pure roll, and coupled sway-roll excitations, to gain
insights into the behaviour of the liquid and to understand
the potential impacts on the structural integrity of the tank
under these specific conditions.
The model considered in the study is a 1:50 scaled
membrane-type tank based on the design of a 48,280 m3 KC-1
membrane LNG tank designed by Korea Gas Corporation
(KOGAS) [6]. To investigate the sloshing loads, the variations
Fig. 1. An example of the distribution of 𝛼𝛼 values in cells [8]
in the excitation frequency and amplitude are considered
in order to observe the time-dependent sloshing pressure Fig. 1. An example of the distribution of α values in cells [8]
variations over a wide range of sway, roll, and coupled EQUATIONS
GOVERNING sway-
roll excitations, using the Ansys Fluent computational fluid
dynamics (CFD) software. MATLAB software is also used
In this study, a conventional
to perform a fast Fourier transform (FFT) analysis of the GOVERNING EQUATIONS
numerical approach is employed to calcula
phenomenon of
time series results for the sloshing pressures under coupled liquid in a tank. It is assumed that the fluid flow is incompressible, a
sway-roll excitations, to provide aflow in the sloshing
quantitative analysis inphenomenon is represented by the Reynolds-averaged Navier-
the frequency domain. equation. The Reynolds stresses In thisand study,theaturbulent
conventional flownumerical
field in the approach
RANSisare calc
employed to calculate the
the standard k-ε model. The volume fraction for the two phases, liquid sloshing phenomenon of liquid
and in gas, is
a
the volume fraction equation in Eq. (2): tank. It is assumed that the fluid flow is incompressible, and
the turbulent flow in the sloshing phenomenon is represented
1 𝜕𝜕by the Reynolds-averaged Navier-Stokes (RANS)
[ (𝛼𝛼𝑞𝑞 𝜌𝜌𝑞𝑞 ) + 𝛻𝛻 ∙ (𝛼𝛼𝑞𝑞 𝜌𝜌𝑞𝑞 𝑢𝑢 ⃗ 𝑞𝑞 ) = 𝑆𝑆𝛼𝛼𝑞𝑞 + ∑𝑛𝑛𝑝𝑝=1(𝑚𝑚̇equation.
𝑝𝑝𝑝𝑝 − 𝑚𝑚̇𝑞𝑞𝑞𝑞 )]
𝜌𝜌𝑞𝑞 𝜕𝜕𝜕𝜕The Reynolds stresses and the turbulent flow field in the RANS
are calculated based on the standard k-ε model. The volume
where the variables 𝑚𝑚̇𝑝𝑝𝑝𝑝 and 𝑚𝑚̇𝑞𝑞𝑞𝑞 represent the mass transfer from phases 𝑝𝑝 to 𝑞𝑞 an
to 𝑝𝑝, respectively [9, 10]. 𝑆𝑆𝛼𝛼𝑞𝑞 is the source term for phase 𝑞𝑞 in a cell [9, 10]. Eq. (3)
44 equation for
POLISH MARITIME RESEARCH, No the momentum, as follows:
4/2023
𝜕𝜕
(𝜌𝜌𝑢𝑢
⃗ ) + 𝛻𝛻 ∙ (𝜌𝜌𝑢𝑢 ⃗ ) = −𝛻𝛻𝛻𝛻 + 𝛻𝛻 ∙ (𝜏𝜏𝜏𝜏) + 𝜌𝜌𝑔𝑔 + 𝐹𝐹
⃗ 𝑢𝑢
𝜕𝜕𝜕𝜕
where 𝑢𝑢
⃗ denotes the velocity vector, 𝜌𝜌 the density of the mixture, 𝑝𝑝 the static pre
excitation induces a linear transverse (lateral
𝑋𝑋 andor side-to-side) themotion along the tank’s tran
expression
displacement, for the roll
respectively. 𝑋𝑋0 represents
excitation. Here, the 𝜃𝜃amplitude
represent of lateral
the displacement
lateral and
displacement, 𝜃𝜃
th
whereas roll excitation induces
𝑋𝑋 a tilting rotation of the of tank about itsdisplacement,
longitudinal ax𝜃𝜃
example
An of the
example distribution
of the of 𝛼𝛼
distribution Fig.𝛼𝛼 1.
ofvalues Aninexample
values cells [8]of
in cells
displacement,
[8]the distribution ofthe amplitude
𝛼𝛼 values in cells [8] of the rotational0 displacement, and 𝜔𝜔 is the frequency of the excitations. 0
respectively. represents the amplitude the lateral
centre
the of the bottom
amplitude of the tank
of the rotational acting as theand
displacement, centre
𝜔𝜔 isofthe
thefrequency
coordinate of system. If the sw
the excitations.
NING
the distribution of 𝛼𝛼 values
EQUATIONS in An
Fig. 1. cells excitations
[8] of the distribution of 𝛼𝛼 values in cells [8]
example
are periodic, they can be represented in the form of sinusoidal waves, as sho
G
ofvaluesEQUATIONS
𝛼𝛼study, in cellsin a
[8]conventional numerical
𝑋𝑋 = 𝑋𝑋 𝑠𝑠𝑠𝑠𝑠𝑠 (𝜔𝜔𝜔𝜔) 0
numerical approach is isemployed toapproach
tocalculate istheemployed to calculate the sloshing
issloshing
values
numerical cells [8]
approach employed calculate the sloshing
EQUATIONS
non of liquid fraction in a for the
tank. It two
is assumed phases,
Fig. 1. An thatliquid
example the of
and
fluid
the
gas,
flow
distribution
calculated
isof 𝛼𝛼
incompressible,
values in
using
cells [8] and the turbulent
assumed
is assumed that that thethe fluid fluid flow flow is is incompressible,
incompressible, and and thethe turbulent
turbulent 𝜃𝜃 = 𝜃𝜃0 𝑠𝑠𝑠𝑠𝑠𝑠 (𝜔𝜔𝜔𝜔)
he
NING
sloshing thephenomenon
EQUATIONS volume fraction is represented equationby inthe Eq.Reynolds-averaged
(2): Navier-Stokes (RANS) (6)
al
is
dy, approach
represented
represented
a conventional byisby the employed
the Reynolds-averaged
Reynolds-averaged
numerical to calculate approach theis sloshing
Navier-Stokes
Navier-Stokes employed (RANS)
(RANS)to calculate the sloshing
nd
of The
d,GOVERNING
that
theathe
liquid Reynolds
the
turbulent influid
turbulent
conventional a tank. stresses
flow
flow EQUATIONS
flow is field
field
Itnumerical and
incompressible,
assumed in in thethe
the turbulent
RANS
that
approach RANStheand are
fluid flow
are
is the
calculated field
turbulent
calculated
flow
employed inbased the
based
is incompressible,
to is RANS onon areand
calculate calculated
the thesloshing
turbulent based on
sard
me nted study,
fraction k-ε by
fraction model.
the a for
for conventional
The
Reynolds-averaged
the the two volume
two phases, phases, numerical
fraction
liquid liquid for
Navier-Stokes
and approach
and the
gas, gas,two
is
Eq.
is
phases,
(RANS)
calculated
is calculated
(5)
employed liquid
using using
the to mathematical
and calculate
gas, is calculated expression
theEq.sloshing
(5)
using for the swayexpression
is the mathematical excitation, forwhile Eq. (6) is the m
the sway
oshing
iquid
is employed
h isofemployed inphenomenon
a tank. toIt is is
assumed
calculate
to calculate represented that
the the by
sloshing
the sloshing the
fluid Reynolds-averaged
flow is incompressible, Navier-Stokes
and the (RANS)
turbulent
me non
bulent
duid2): fraction
.Reynolds
ing(2):
flowIn
liquid
flow
phenomenon
flow is
this is stresses
in
equation
field in
incompressible,
study,
a
incompressible,
tank.
ain
isand the It
Eq.
represented
is
RANS
the and
conventional
assumed
(2):
turbulent
thethe are
by
that
calculated
the
turbulent flow
numerical
the fluid
fieldexpression
Reynolds-averaged based
in flow
theonRANS
approach
is incompressible,
for are
isNavier-Stokes the
employed roll to
calculated (2) and excitation,
the
excitation.
based
(RANS)
turbulent
calculate on the while
Here, 𝑋𝑋 and 𝜃𝜃 represent the lateral displacement and th
Eq.
sloshing
(6) is the mathematical expression
is and turbulent
for using
the roll excitation. X and θ of
theHere, represent the lateral
he sloshing phenomenon represented by the Reynolds-averaged Navier-Stokes (RANS)
neynolds
-ε formodel. the stresses
eynolds-averaged
phenomenon
Reynolds-averaged twoThe 1 phases,
[
volume
of and
𝜕𝜕 liquid
(𝛼𝛼
liquid
the
Navier-Stokes
𝜌𝜌
fraction
in
Navier-Stokes
) +
a
𝑛𝑛 𝑛𝑛
andfor
turbulent
tank.
𝛻𝛻 ∙ (𝛼𝛼
gas,
It the
flow
(RANS)
is
𝜌𝜌
is
assumed
(RANS)
𝑢𝑢
⃗
calculated
two
)
field
=
phases,
𝑆𝑆
inthat theusing
displacement,
+ ∑
liquid
the RANS fluid
(𝑚𝑚
and
̇
are
flow gas,
𝑛𝑛in the RANS are calculated based
−
respectively.
calculated
is
𝑚𝑚 ̇
is calculated
incompressible,
) ]
based 𝑋𝑋
on represents
and
0 (2) the turbulent amplitude the lateral displacement, 𝜃𝜃
+𝛻𝛻 The
waction
model.
ield ∙
𝛻𝛻(𝛼𝛼∙inin 𝜌𝜌
Reynolds
(𝛼𝛼
𝑞𝑞equation
the 𝑞𝑞𝑞𝑞
The where
𝑢𝑢
⃗
𝜌𝜌 𝑢𝑢
⃗) =
)
𝑞𝑞𝑞𝑞sloshing
𝑞𝑞RANS the
𝑆𝑆
stresses
=
𝑞𝑞𝜕𝜕𝜕𝜕in 𝛼𝛼
volume Eq.𝑆𝑆
are𝑞𝑞variables
𝑞𝑞𝛼𝛼
+ 𝑞𝑞
(2):
fraction ∑
+ and ∑
𝑝𝑝=1
calculated (𝑚𝑚
the
𝑝𝑝=1 ̇
(𝑚𝑚
for𝑝𝑝𝑝𝑝 ̇
𝑞𝑞
the and
−
turbulent
𝑝𝑝𝑝𝑝
based 𝑞𝑞−𝑚𝑚
two ̇
𝑞𝑞 𝑚𝑚
𝑞𝑞𝑞𝑞̇ ) ]
flow
𝑞𝑞𝑞𝑞
onphases, represent
) ]𝛼𝛼 field
𝑞𝑞 liquid 𝑝𝑝=1 the
and mass (2)
𝑝𝑝𝑝𝑝 (2) transfer
𝑞𝑞𝑞𝑞
gas, is calculatedNavier-Stokes displacement
using
on and the rotational displacement, respectively.
flow
field
ard k-ε model. in the
the 𝜌𝜌 RANSThe volume phenomenon
are
𝑞𝑞 calculated
pistostresses fraction is represented
based
forphases on
the two qthe
by the
pamplitude
phases, Reynolds-averaged
liquid and of[9,gas,the rotational
is calculated displacement, (RANS)
using the amplitude and 𝜔𝜔ofisthe thelateral
frequency of the excitations.
ion
phases,
oequation.equation
phases, 1 𝜕𝜕
from
liquid The
liquid phases
inand Eq.
Reynolds
and 𝑛𝑛
(2):
gas, gas, isq(2):and from
calculated
calculated andusing theusing turbulent to ,flow
respectivelyfield in the 10].
RANS are X0 represents
calculated based on displacement,
me𝜌𝜌 𝑞𝑞 𝑢𝑢
⃗ 𝑞𝑞standard
variables
present
represent
the ) =the
fraction [𝑆𝑆the 𝑚𝑚
𝛼𝛼mass +
equation
̇ 𝑝𝑝𝑝𝑝
(𝛼𝛼 ∑
𝜌𝜌
is𝑞𝑞and
mass the
transfer
𝑞𝑞𝑝𝑝=1 𝑚𝑚in(𝑚𝑚
)transfer
+ source 𝛻𝛻̇ 𝑝𝑝𝑝𝑝
̇ 𝑞𝑞𝑞𝑞 Eq. from −term
𝑞𝑞 𝑚𝑚
∙represent
(𝛼𝛼from 𝑞𝑞̇ 𝑞𝑞𝑞𝑞
𝜌𝜌phases 𝑢𝑢 𝑞𝑞)the
⃗phases
for)] =phase 𝑆𝑆to
𝑝𝑝mass 𝑝𝑝𝛼𝛼𝑞𝑞to 𝑞𝑞∑
𝑞𝑞+transfer
qand in
and
𝑛𝑛
athe
from
𝑝𝑝=1 (𝑚𝑚
cell
(2)
from froṁ phases
[9,
𝑝𝑝𝑝𝑝 − 10].
phases
phases 𝑞𝑞)𝑝𝑝
𝑚𝑚̇𝑞𝑞𝑞𝑞𝑞𝑞Eq. to 𝑞𝑞isand
]liquid
(3) andfrom θgas, represents
phases
(2) 𝑞𝑞 the using
amplitude of the rotational displacement,
1 𝜌𝜌𝑞𝑞𝜕𝜕 𝜕𝜕𝜕𝜕 𝑞𝑞 k-ε model. The volume fraction 𝑛𝑛
for two phases, 0 is calculated
ectively
eource [
term [9, (𝛼𝛼
the 1
10].
𝑞𝑞
for 𝜌𝜌 𝜕𝜕
𝑞𝑞 )
governing
𝑆𝑆 + 𝛻𝛻𝑞𝑞 in
is ∙the(𝛼𝛼
𝑞𝑞 in 𝜌𝜌
equation
source
𝑞𝑞 𝑞𝑞 𝑢𝑢
⃗ 𝑞𝑞 ) =for
term 𝑆𝑆 the
𝛼𝛼 for + ∑
momentum,
phase 𝑝𝑝=1 (𝑚𝑚
𝑞𝑞 iṅ 𝑝𝑝𝑝𝑝
𝑛𝑛 a − as
cell 𝑚𝑚 ̇ follows:
[9,
𝑞𝑞𝑞𝑞 ) ]
10]. Eq. (3) is and
the (2) ω is
governing the frequency of the excitations.
the
source volume
𝜌𝜌𝑛𝑛
∑mass 𝑞𝑞 𝜕𝜕𝜕𝜕
term [ phase
fraction
for phase
(𝛼𝛼
𝛼𝛼𝑞𝑞 𝜌𝜌 equation a cell
a cell
𝑞𝑞)]𝑞𝑞 ) + 𝛻𝛻 ∙ (𝛼𝛼𝑞𝑞 𝜌𝜌𝑞𝑞 𝑢𝑢
in[9, Eq. [9,10]. ⃗ 𝑞𝑞𝑞𝑞Eq.
(2):
10]. ) Eq. =(3) 𝑆𝑆𝛼𝛼(3) is+isthe ∑the governing
governing
𝑝𝑝=1(𝑚𝑚̇𝑝𝑝𝑝𝑝 − 𝑚𝑚̇𝑞𝑞𝑞𝑞 )] (2)
+ the
ables
𝛼𝛼for + ∑
𝑝𝑝=1 𝑚𝑚𝑛𝑛̇ (𝑚𝑚
𝑝𝑝𝑝𝑝
̇
(𝑚𝑚 𝜌𝜌𝑞𝑞̇ −
transfer
and
𝑝𝑝𝑝𝑝 𝑚𝑚
𝜕𝜕𝜕𝜕
𝑝𝑝𝑝𝑝
𝑚𝑚
− ̇ ̇
from
𝑞𝑞𝑞𝑞 𝑚𝑚
𝑞𝑞𝑞𝑞 ̇ represent
𝑞𝑞𝑞𝑞 ) ]
phases 𝑝𝑝
the to mass𝑞𝑞 (2)
and from
transfer
(2) 𝑞𝑞phases
from 𝑞𝑞
phases 𝑝𝑝 to 𝑞𝑞 and from Fig.
phases 𝑞𝑞
2 shows the two-dimensional geometric parameters
ws:𝑞𝑞 the𝑝𝑝=1
lows: momentum, as1 follows: 𝜕𝜕 Fig. 2. Parameters for a two-dimensional rectangular tank (left) and a two-dimensional prismatic tank (right).
∑ 𝑛𝑛
ermles
ely 𝑚𝑚
for
[9, ̇ 𝑝𝑝𝑝𝑝10]. and𝑆𝑆𝛼𝛼𝑚𝑚𝑞𝑞̇ 𝑞𝑞𝑞𝑞
phase in
is represent
a
the cell [[9,(𝛼𝛼
source the
10]. 𝑞𝑞
term 𝜌𝜌massEq.
𝑞𝑞 )for +(3) 𝛻𝛻phase∙is(𝛼𝛼
transfer the𝑞𝑞𝑞𝑞𝜌𝜌from 𝑢𝑢
⃗ a𝑞𝑞 )cell
governing
𝑞𝑞
in = 𝑆𝑆[9,
phases 𝛼𝛼 +
𝑝𝑝
10]. to 𝑞𝑞
𝑝𝑝=1
Eq. (𝑚𝑚
and (3) ̇ from
𝑝𝑝𝑝𝑝
is −
the 𝑚𝑚 ̇
phasesof
𝑞𝑞𝑞𝑞
governing)a] 𝑞𝑞
rectangular tank
(2) and a prismatic model tank. prismatic
The tank (right).
e variables 𝑚𝑚̇𝑝𝑝𝑝𝑝𝑞𝑞 and 𝑚𝑚
𝜌𝜌 ̇ 𝑞𝑞 𝜕𝜕𝜕𝜕
represent
𝜕𝜕 to 𝑞𝑞𝑞𝑞𝑞𝑞and from phases 𝑞𝑞 the mass transfer from 𝑞𝑞
phases Fig. 𝑝𝑝
2. to 𝑞𝑞
Parameters and for
from a phases 𝑞𝑞
two-dimensional rectangular tank (left) and a two-dimensional
+sfer
𝛻𝛻ectively from
(𝜌𝜌𝑢𝑢
∙momentum,
𝛻𝛻 (𝜌𝜌𝑢𝑢
∙ ⃗ 𝑢𝑢 )
⃗⃗ [9, 𝑆𝑆
phases
⃗=
𝑢𝑢 )
𝛼𝛼𝑞𝑞= −𝛻𝛻𝛻𝛻−𝛻𝛻𝛻𝛻 𝑝𝑝 ++ 𝑝𝑝
(𝜌𝜌𝑢𝑢 𝛻𝛻 ⃗ ) 𝑞𝑞
+
(𝜏𝜏𝜏𝜏 ) 𝛻𝛻
(𝜏𝜏𝜏𝜏
∙the∙ source )
++ ∙ (𝜌𝜌𝑢𝑢
𝜌𝜌𝑔𝑔𝜌𝜌𝑔𝑔 ⃗ ++𝑢𝑢
⃗ ) = −𝛻𝛻𝛻𝛻𝑞𝑞 𝑞𝑞 +
𝐹𝐹 𝐹𝐹for phase 𝑞𝑞 in a cell (3) 𝛻𝛻 ∙ (𝜏𝜏𝜏𝜏 ) + 𝜌𝜌𝑔𝑔 + 𝐹𝐹 rectangular tank is considered for the validation of modelling
is 𝛻𝛻
yansfer[9, 10]. from phases
is the sourceto and
term from for phases
phase in a cell [9, 10]. Eq. (3) is the
(3) governing (3)
he 10]. as𝑆𝑆𝜕𝜕𝜕𝜕𝛼𝛼 follows: term Fig.transfer 2 shows [9,(3) 10]. Eq. the (3) is𝑝𝑝 the
to 𝑞𝑞governing
two-dimensional geometric parameters
𝑞𝑞where
se 𝑞𝑞 in
in
momentum, a cell athe cell variables
[9, [9,
as 10]. 10].
follows: 𝑞𝑞 𝑚𝑚̇(3) and
Eq. Eq. 𝑝𝑝𝑝𝑝(3) is theis𝑚𝑚 the ̇ governing represent
𝑞𝑞𝑞𝑞governing the mass from phases assumptions
and fromused phases in 𝑞𝑞the computation, andof theamembrane-
rectangular tank and
𝜕𝜕 Fig. 2 shows the two-dimensional geometric parameters of a rectangular
for
𝑢𝑢
⃗,or,)𝜌𝜌=
denotes
to the
𝜌𝜌 −𝛻𝛻𝛻𝛻
𝑝𝑝,
the the momentum,
the where
density
respectively +velocity
density ∙of(𝜏𝜏𝜏𝜏
𝛻𝛻(𝜌𝜌𝑢𝑢 ⃗ of )[9,
the
𝛼𝛼𝑞𝑞
as
)denotes
++ follows:
vector,
the 𝜌𝜌𝑔𝑔
10]. (𝜌𝜌𝑢𝑢
𝛻𝛻mixture,
∙mixture, 𝜌𝜌𝐹𝐹𝑢𝑢
𝑆𝑆+⃗the )𝑝𝑝the
the
⃗ is = 𝑝𝑝density
velocity
the −𝛻𝛻𝛻𝛻
the static
source +vector,
static of
𝛻𝛻 model
(𝜏𝜏𝜏𝜏for
∙the
pressure,
term ρ(3)
)mixture,
pressure, + the
𝜌𝜌𝑔𝑔
phase tank.
and +
anddensity 𝑝𝑝the
𝜏𝜏𝜏𝜏𝑞𝑞𝐹𝐹 𝜏𝜏𝜏𝜏in
theThe
the ofrectangular
static
a cell the 10]. type
[9,pressure, Eq. (3) and
(3) is𝜏𝜏𝜏𝜏the
tank
tank isgoverning
with
the considered
a prismaticfor the isvalidation
shape the modelof tank tank
modelling
LNG assumptand
𝜕𝜕 𝜕𝜕𝜕𝜕
sor
present
epresent
equation[11,the 12].
the
for (𝜌𝜌𝑢𝑢
mixture, 𝜌𝜌𝑔𝑔 +
gravitationalandp𝛻𝛻the
⃗ ) 𝜕𝜕momentum,
gravitational
the ∙𝐹𝐹(𝜌𝜌𝑢𝑢
static
⃗ 𝑢𝑢
represent
body ⃗)=
body pressure,
force−𝛻𝛻𝛻𝛻
theand
force + and
gravitational
and (𝜏𝜏𝜏𝜏model
𝛻𝛻external )the
∙ external
the + 𝜌𝜌𝑔𝑔stress tank.
+
computation,
bodybody
body tensor
𝐹𝐹force,
forceThe
force, [11,
andrectangular
12].
and theused
external (3)tank theisstudy.
inforce,
membrane-type
body considered for the
tank with validationshape
a prismatic of modelling
is the model assumptiLNG
(𝜏𝜏𝜏𝜏 )
𝜕𝜕𝜕𝜕 (𝜌𝜌𝑢𝑢⃗ )𝜌𝜌𝑝𝑝+the 𝛻𝛻 ∙as(𝜌𝜌𝑢𝑢 follows:
⃗ )pressure,
⃗ 𝑢𝑢 = −𝛻𝛻𝛻𝛻 +and 𝛻𝛻 ∙𝜏𝜏𝜏𝜏(𝜏𝜏𝜏𝜏the
)𝑝𝑝+the 𝜌𝜌𝑔𝑔static
+ 𝐹𝐹 pressure, and 𝜏𝜏𝜏𝜏 the (3)
𝛻𝛻
+ ∙
density
es 𝛻𝛻 the
∙ +
(𝜏𝜏𝜏𝜏
or, 𝜏𝜏𝜏𝜏, 𝜏𝜏𝜏𝜏can
ely
nsor, [11, )
, canof 𝜌𝜌𝑔𝑔
velocity
+
12]. the
bebe +
and
𝜌𝜌𝑔𝑔
The 𝐹𝐹
mixture,
+vector,
𝐹𝐹
𝜕𝜕𝜕𝜕
expressedstress
expressed represent
tensor,
as:as: the the
static
𝜏𝜏𝜏𝜏
density
, gravitational
can be of
(3) the
expressed
(3) the body
mixture, force
computation,
as:
in the study. and external
and the To model
membrane-type the sway and
tank roll
with excitations
a prismatic of the
shapetank,is the
the model LNG
𝜌𝜌 𝜕𝜕density of the mixture, 𝑝𝑝 the static pressure, and 𝜏𝜏𝜏𝜏 the
thegravitational
the
[11, velocity
12]. body force,
vector,
𝜌𝜌𝑔𝑔velocity
and 𝐹𝐹body respectively
representthe
force (𝜌𝜌𝑢𝑢
𝜌𝜌and
the⃗ ) gravitational
+[11,
external ∙ (𝜌𝜌𝑢𝑢
𝛻𝛻 12]. ⃗of⃗ )in
The
𝑢𝑢
body stress
=
body−𝛻𝛻𝛻𝛻
force,
the tensor,
+ 𝛻𝛻𝑝𝑝and
force
study. (𝜏𝜏𝜏𝜏),external
∙ the +can𝜌𝜌𝑔𝑔be+ 𝐹𝐹
bodyfirstforce,
natural thefrequency
𝜏𝜏𝜏𝜏(4) (3) of the prismatic tank was used. The
denotes
ethe 𝜏𝜏𝜏𝜏expressed
mixture, = the
𝜏𝜏𝜏𝜏𝜌𝜌𝑔𝑔
=𝜇𝜇(𝛻𝛻𝑢𝑢 ⃗𝑝𝑝 +
𝑝𝑝𝜇𝜇(𝛻𝛻𝑢𝑢
the ⃗the
𝐹𝐹 +𝛻𝛻𝑢𝑢
static 𝑇𝑇vector,
⃗𝛻𝛻𝑢𝑢 𝑇𝑇
⃗) pressure, 𝜕𝜕𝜕𝜕
) pressure, the
𝜏𝜏𝜏𝜏anddensity
= 𝜇𝜇(𝛻𝛻𝑢𝑢 𝜏𝜏𝜏𝜏⃗ the
𝜏𝜏𝜏𝜏 the + 𝛻𝛻𝑢𝑢 ⃗ 𝑇𝑇 )mixture,
the (4)(4) static pressure, and
1,
an
11, mixture,
12].
be
12]. Theexpressed
and stress as: as:
static
represent
tensor, 𝜏𝜏𝜏𝜏,the be and
cangravitational
expressed as: To
body force model and the
external sway body andfirst
force, natural
roll frequencyofofthe
excitations thetank,
rectangular
the firsttank, ω1, canfrequency
natural be of th
nsor [11,𝑢𝑢 ⃗ 12]. 𝜌𝜌𝑔𝑔 and 𝐹𝐹 represent the 𝜌𝜌gravitational body force and 𝑝𝑝external body force, and 𝜏𝜏𝜏𝜏 the
where
nal
12].
tional body The
body denotes
force
stressforce and the
tensor,
and 𝜏𝜏𝜏𝜏
velocity
external ,
externalcan vector,
body
be force,
expressed
body force, the density
as: To
tankas: of
model the mixture,
the
was used. The first natural sway the determined
static
and pressure,
roll using
excitations Eq. (7),
theand its
tank, correlation
the first with the
of the rectangular tank, 𝜔𝜔1 , can be determine
of natural first
frequency of th
(4) frequency
s, 12].
11, the 12]. dynamic 𝑇𝑇 )viscosity [11, 12].
=ed
ely𝜇𝜇(𝛻𝛻𝑢𝑢
[11, ⃗ tensor
+12].𝛻𝛻𝑢𝑢⃗ The [11, 12]. 𝜌𝜌𝑔𝑔𝜏𝜏𝜏𝜏and
stress tensor, = 𝜏𝜏𝜏𝜏𝜇𝜇(𝛻𝛻𝑢𝑢
, can⃗ be+ 𝛻𝛻𝑢𝑢
𝐹𝐹 represent ⃗ 𝑇𝑇 )the
expressed (4)
ssed as:as:
stress gravitational body force andnatural
externalfrequency
body of
force,the prismatic tank, ω' , is expressed
𝜔𝜔1 , can in ′
𝜏𝜏𝜏𝜏 = 𝜇𝜇(𝛻𝛻𝑢𝑢 ⃗ + 𝛻𝛻𝑢𝑢 ⃗ 𝑇𝑇 ) tank (7), andwasits used. The firstwith
correlation natural frequency
(4)the first naturaloffrequency
the rectangular of the tank,
prismatic
1 tank,be𝜔𝜔determine
1 , is expr
𝑇𝑇respectively [11, 12]. The stress𝜏𝜏𝜏𝜏 tensor,
= 𝜇𝜇(𝛻𝛻𝑢𝑢 ⃗ 𝜏𝜏𝜏𝜏,+can 𝛻𝛻𝑢𝑢⃗ 𝑇𝑇 )expressed
be as: (4) Eq. (8) [13–15]:
(4) ′
dynamic
⃗) )
𝑢𝑢 𝑇𝑇 viscosity [11, 12]. (4)(4) (7), and its correlation
(8) [13–15]: Fig. 2. Parameters with for the first natural
a two-dimensional frequency
rectangular ofand
tank (left) thea two-dimensional
prismatic tank, 𝜔𝜔1tank
prismatic , is(right).
expre
ynamic viscosity [11, 12].
s the dynamic viscosity [11, 12]. 𝜏𝜏𝜏𝜏 = (8)𝜇𝜇(𝛻𝛻𝑢𝑢 ⃗ + 𝛻𝛻𝑢𝑢
[13–15]: ⃗ 𝑇𝑇 ) (4)
where μ is the dynamic viscosity [11, 12]. 𝜋𝜋𝜋𝜋 𝜋𝜋ℎ
where 𝜇𝜇 is the dynamic viscosity [11, 12]. Fig. 2 shows the two-dimensional geometric 𝜔𝜔12 = 𝜋𝜋𝜋𝜋 𝑡𝑡𝑡𝑡𝑡𝑡ℎ (𝜋𝜋ℎ ) of a rectangular
parameters (7) tank and
𝐵𝐵 𝐵𝐵
model tank. The rectangular tank is considered 𝜔𝜔12 = for 𝑡𝑡𝑡𝑡𝑡𝑡ℎ
the (validation) of modelling assumpti
𝐵𝐵 𝐵𝐵
EXCITATION FORCES the computation, and the membrane-type 𝜔𝜔1′
2
[𝐵𝐵1tank−1
𝐻𝐻1 𝑠𝑠𝑠𝑠𝑠𝑠ℎ(
𝜋𝜋𝐻𝐻1 2 (𝜋𝜋𝐵𝐵1 )]
)−𝐵𝐵1 𝐻𝐻1−1 𝑠𝑠𝑠𝑠𝑠𝑠shape
with𝜋𝜋𝐻𝐻𝐵𝐵a prismatic 𝐵𝐵 is the model LNG
𝜔𝜔 ′ 22 = 1 − [𝐵𝐵1 𝐻𝐻1−1 𝑠𝑠𝑠𝑠𝑠𝑠ℎ( 1 )−𝐵𝐵2𝜋𝜋ℎ
𝐻𝐻 −1 𝑠𝑠𝑠𝑠𝑠𝑠2 (𝜋𝜋𝐵𝐵1 )]
in the study. 𝜔𝜔11 𝜋𝜋𝜋𝜋𝜋𝜋𝜋𝜋ℎ(
𝐵𝐵 1 1
) 𝐵𝐵
=1− 𝐵𝐵
2𝜋𝜋ℎ (8)
𝜔𝜔1 2 𝜋𝜋𝜋𝜋𝜋𝜋𝜋𝜋ℎ( )
The tank can experience external perturbations
To model including
theEq’ns. 𝐵𝐵
sway and roll excitations of the tank, the first natural frequency of th
sway and roll excitations. Sway excitationIn this study,
induces a linear (5) to (8) are used to apply external excitations to the tank, in ord
FORCES tank
In was
this
transverse (lateral or side-to-side) prescribed
motion
used.
study,
along motionsThe
the Eq’ns.
first(5)natural
tank’saccording to
In(8) tofrequency
this are used
study,
specific ofapply
the rectangular
toformulae.
Eq’ns. (5) to external
(8) To
tank, 𝜔𝜔external
excitations
are accurately
used 1 ,to
to applysimulate
can
thebe determine
tank,
the in ord
motion
′
(7),
transverse axis, whereas roll excitation and
prescribed
under its motions
induces
these correlation
a tilting
excitations, with
according thetofirst
aexcitations
user-defined natural
to the
specifictank, frequency
in order
formulae.
function (UDF) toTo of
induce
and the prismatic
aprescribed
accurately
dynamic motions
simulate
mesh 𝜔𝜔1 , motion
tank,technique
the is expre
ar
rotation of the tank about its longitudinal
experience external perturbations including (8)
under axis, with
[13–15]:
thesesway
to impose the centre
excitations,
the and roll
external according
a user-defined
excitations.
excitation to specific
functionformulae. To accurately
Sway (UDF) and a dynamic mesh technique ar
effectively. simulate the
of the bottom of the tank acting as thetocentre
impose of the
thecoordinate
external motion ofeffectively.
the tank under these excitations, a user-defined
s a linearsystem.
transverse (lateral or side-to-side) motion along
If the sway and roll excitations are periodic, they can
the excitation
tank’s transverse axis,
function (UDF) and 𝜋𝜋𝜋𝜋 mesh technique𝜋𝜋ℎ
itation induces a tilting rotation of the tank about its longitudinal axis, with 𝜔𝜔athe
2dynamic
1 = 𝐵𝐵 𝑡𝑡𝑡𝑡𝑡𝑡ℎ ( 𝐵𝐵 )
are employed
be represented in the form of sinusoidal waves, as shown to impose the external excitation effectively.
om of the tank acting as the centre of the coordinate system. If The
below. the ′investigation
sway
2 and rollfocused on the effects of𝜋𝜋𝐵𝐵different
𝜋𝜋𝐻𝐻
[𝐵𝐵1 𝐻𝐻1−1 𝑠𝑠𝑠𝑠𝑠𝑠ℎ( 𝐵𝐵 1 )−𝐵𝐵1 𝐻𝐻1−1 𝑠𝑠𝑠𝑠𝑠𝑠2 ( 𝐵𝐵 1 )]
𝜔𝜔1 on the sloshing
excitations loads in the tank, including single
2 = 1−
riodic, they can be represented in the form of sinusoidal waves, as shown below. 2𝜋𝜋ℎ
𝜔𝜔1
excitations (pure sway and pure roll) and
𝜋𝜋𝜋𝜋𝜋𝜋𝜋𝜋ℎ( ) multiple coupled
𝐵𝐵
𝑋𝑋 = 𝑋𝑋0 𝑠𝑠𝑠𝑠𝑠𝑠 (𝜔𝜔𝜔𝜔) excitations '(coupled
(5)sway-roll)'.
(5) In particular, the sloshing
pressures
In this study, Eq’ns. (5) in the
to (8) are tanktoare
used analysed
apply when the
external excitationto
excitations of roll
the tank, in ord
𝜃𝜃 = 𝜃𝜃0 𝑠𝑠𝑠𝑠𝑠𝑠 (𝜔𝜔𝜔𝜔)
prescribed motions according to specific(6)formulae. To accurately simulate the motion
under these excitations, a user-defined function (UDF) and a dynamic mesh technique ar
mathematical expression for the sway excitation, while Eq. (6) is the mathematical
to impose the external excitation effectively.
e roll excitation. Here, 𝑋𝑋 and 𝜃𝜃 represent the lateral displacement and the rotational
spectively. 𝑋𝑋0 represents the amplitude of the lateral displacement, 𝜃𝜃0 represents
the rotational displacement, and 𝜔𝜔 is the frequency of the excitations.
Fig. 2. Parameters for a two-dimensional rectangular tank (left) and a two-dimensional prismatic tank (right)
Parameters for a two-dimensional rectangular tank (left) and a two-dimensional prismatic tank (right).
is coupled with the sway’s in the same-phase (0°) and the In the numerical simulation, the fluid motion was
opposite-phase (180°). modelled as an incompressible unsteady flow, with a no-slip
Fig. 3 shows the time-dependent lateral displacements of condition assumed at the walls. To track the free surface,
a tank subjected to a pure sway excitation having an amplitude which represents the interface between the water and air,
of 0.015 m, while others are the time-dependent rotational the VOF method was employed [7]. To express the turbulent
displacements due to pure roll excitations with amplitudes of flow of the liquid, the standard k-ε model was used [17]. To
1°, 3°, and 5° in the same-phase or the opposite-phase relative calculate the pressure on a computational mesh from the
to the sway’s. The frequency of the excitations was set to the velocity components, the semi-implicit method for pressure-
first natural frequency of the tank, denoted as 1.0 ω'1. link equations (SIMPLE) algorithm was used, in which the
momentum equations were
coupled with an iterative
procedure [18]. A compressive
scheme was used to calculate
the volume fraction for the
interface capturing problem [19].
Fig. 4 illustrates the base
case for the computational
mesh used in the simulation
of the rectangular tank, which
was formed of a combination
of triangular and rectangular
grids. When excitation is
applied to a tank, sloshing-
induced liquid forces often
impacts and shocks certain
zones on the upper-left and
upper-right walls of the tank,
and on the roof. Furthermore,
Fig. 3. Displacement of a tank due to pure sway and pure roll excitations. The the motion of the liquid near the point of impact exhibits
roll excitation is in-phase or out-of-phase with the sway excitation complex flow phenomena [20]. Thus, to accurately capture
the complex flow behaviour of the liquid, smaller triangular
grids (0.0025 m here in the base case) were employed near
VALIDATION the anticipated impact zones on the walls [20, 21]. Larger
rectangular grids (0.005 m in the base case) were used in the
remaining regions to reduce the computational time while
To assess the validity of the modelling assumptions used still maintaining reasonable accuracy. The entire mesh for
for the numerical simulation, a comparison was conducted the base case consisted of a total of 64,620 grid points. To
between the simulation results and experimental data obtained assess the numerical accuracy of the results in terms of the
from a well-known prior study conducted by Hinatsu [16]. sloshing loads in the tank with respect to the grid resolution,
The experimental data selected for the comparison consisted two cases were analysed: the base case and a coarse case of
of measurements of sloshing pressures in a rectangular tank 32,132 grid points, half the number of the base case. It is
under pure sway excitation. A schematic representation of important to note that in both cases, the number of grid points
the rectangular tank used in their experimental investigation significantly exceeded those used in a validation process
is presented in Fig. 4. The tank had a base length of 1.2 (7,200) performed by Rhee [12]. Furthermore, our study
m and height of 0.6 m. For the experiment, the tank was investigated the influence of the time step size on accuracy.
filled with water to 60% of its height from the floor, and the Three time steps were compared: a base time step of 0.0005 s,
remaining spaces were filled with air. Subsequently, the tank a large time step of 0.001 s (twice that of the base case), and an
was subjected to pure sway excitation characterised by an extra-large time step of 0.002 s (twice that of the large case).
amplitude of 0.015 m and a period of 1.404 s. During the
validation process, the simulation results were compared
to the experimental data pertaining to the sloshing impact
pressure at a specific location, denoted as P1 at the tank
wall (as indicated in Fig. 4). The primary objective of this
comparison was to evaluate the degree of agreement between
our simulation data and the experimental data regarding the
sloshing impact pressures at the designated point.

εvalidation, and the error relative
to the simulation values for the
base cases as ε.
The results for the accuracy
with respect to the number
of grid points showed that ε
was 0.94% for the coarse case,
and the values of εvalidation for
the base and coarse case were
Fig. 4. Schematic of the rectangular tank used in the validation process (left) 0.68% and 1.63%, respectively.
and the base case of the computational mesh for the tank (right) [Units: m] The values for the base case, with a larger number of grid
points, were more consistent with the experimental values.
The results for the accuracy with a varying time step size
gave values for ε of 2.73% and 8.83% for the large and extra-
large cases, respectively; the values of εvalidation for the large
and extra-large cases were 2.06% and 8.21%, respectively,
while εvalidation for the base case was 0.68%. This investigation
revealed that reducing the time step size led to a decrease
in the error. Fig. 6 shows a graph of the variations in
sloshing impact pressure at point P1, and compares the data
obtained from the simulation of the base cases for both the
mesh and time step size with the experimental data. The
Fig. 5. Pressure history at point P1 of the rectangular tank for the base cases experimental data shown in the figure were obtained by
of both the mesh and time step size Fourier decomposition of the raw experimental data [22]. The
graph shows remarkable agreement between our computed
Tab. 1. Sensitivity of sloshing pressure at point P1 of the rectangular tank data and the experimental data, thus confirming that the
to the number of grid points and the time step size between the 20th assumptions used in the simulation were appropriate, and
and 25th cycles the base cases for both the mesh and time step size were
Number of Average pressure therefore adopted for the simulations in the remainder of the
Mesh ε [%] ε validation [%]
grid points [Pa] study. These results validate the reliability and accuracy of
Base 64,620 1,562.317 - 0.68 our numerical simulation in terms of capturing the sloshing
Coarse 32,132 1,577.064 0.94 1.63 behaviour of water in a tank under the specified conditions.
Average pressure
Time step Size [s] ε [%] ε validation [%]
[Pa]
Base 0.0005 1,562.317 - 0.68
Large 0.001 1,519.703 2.73 2.06
Extra-large 0.002 1,424.312 8.83 8.21
Fig. 5 illustrates the variation in sloshing pressure at point P1

of the rectangular tank, obtained from a simulation of the base
cases for both the mesh and the time step size. The sloshing
pressure initially rises, and then goes through a transient
phase where it increases and subsequently decreases. After
the 20th cycle, the pressure variations stabilise, and a steady-
state region is reached.
Table 1 provides assessments of the simulation accuracy Fig. 6. Pressure history at P1 of the rectangular tank between cycles 20 and 25:
based on the number of grid points and the time step size. comparison of the base case simulation vs. the Fourier decomposition of
To evaluate the accuracy, the average values for the top 10% experimental data [22]
of sloshing pressure values at P1 between the 20th and 25th
cycles in the steady-state region were compared with the
experimental values (comparable six cycles in steady-state
region). To investigate the accuracy in terms of the number
of grid points, the time step was set to the base case, whereas
to investigate the accuracy in terms of the time step size,
the mesh was set to the base case. The error between our
simulation data and the experimental data is denoted as

RESULTS AND DISCUSSION It was observed that an increase in the excitation amplitude
led to an increase in the sloshing pressure. This was be
A simulation study was conducted using a prismatic model attributed to the heightened excitation forces acting on the
tank filled with 50% water, as shown in the schematic diagram tank as the amplitude increases, which result in greater
in Fig. 7. The main objective of the investigation was to analyse sloshing-induced forces. However, during pure rolls with
the influence of the excitation frequency and amplitude on frequencies of 0.8 ω'1, 1.2 ω'1, and 1.8 ω'1, it was noted that
the sloshing pressure in the model tank under both sway and the relationship between amplitude and pressure exhibited
roll excitations, beyond the conditions used for the validation a nonlinear softening effect, leading to a reduced rate of
of the rectangular tank. To explore the effect of the excitation increase in the sloshing pressure. In a study conducted by
amplitude, three different sway amplitudes of 0.0075, 0.015, Akyildiz and Ünal [23] on sloshing pressure during pitch
and 0.03 m were analysed. Similarly, three different roll motion, they also found that sloshing pressure exhibited
amplitudes were examined: 1°, 3°, and 5°. In addition, the decreased sensitivity to amplitude, causing a reduction in
excitation was investigated at six different frequencies relative the rate of pressure increase.
to the tank’s primary natural frequency, ω'1, with values of The impact of excitation frequency on sloshing pressure was
0.8 ω'1, 1.0 ω'1, 1.2 ω'1, 1.4 ω'1, 1.6 ω'1, and 1.8 ω'1. To assess the also investigated. Fig. 8 shows that under pure sway excitation,
sloshing pressure under various types of external excitation, the lowest pressure occurred at an excitation frequency of
the excitations were classified into four scenarios: pure sway, 1.6 ω'1, whereas under pure roll excitation, the lowest pressure
pure roll, and coupled sway-roll with both the same-phase occurred at an excitation frequency of 1.4 ω'1, under the same
and the opposite-phase. The study particularly focused amplitude conditions. When the frequency of pure sway or
on analysing the sloshing impact pressures at point P1, as roll was 1.0 ω'1, the highest sloshing pressure was recorded
indicated on the tank wall in Fig. 7. In the same figure, the under the same amplitude conditions. This behaviour was
computational mesh used to simulate the tank is shown, to attributed to resonance, in which the frequency matches the
enable a visualisation of the detailed mesh structure used to tank’s primary natural frequency, causing intense sloshing
capture the sloshing behaviour. In this study, we investigated of the liquid. This resonance effect results in a significant
the effects of sloshing on the tank under various excitation increase in sloshing pressure, and in engineering design,
scenarios with a 50% liquid filling level, as this is a tank filling it is important to avoid the conditions for resonance and
condition known for its severe sloshing-induced damage [5]. their potential detrimental effects on the structural integrity
We conducted a total of 42 simulations, with 18 for pure sway, of tanks [24, 25].
18 for pure roll, and six for coupled sway-roll.
Fig. 7. Schematic of the prismatic model tank (left, actual LNG tank size: H =
Tab. 2. Average pressures at P1 for the prismatic model tank under pure sway
29.71 m, H1 = 4 m, H2 = 16.507 m, H3 = 9.203 m, B = 40.31 m, B1 = 4 m, B2 =
or roll excitations [Units: Pa]
9.203 m, θ1 = θ2 = 45°) and the base case of the computational mesh used for
the tank (right) [Units: m] Pure sway Pure roll
Frequency
Amplitude Amplitude
0.0075 m 0.015 m 0.03 m 1° 3° 5°

PURE SWAY AND PURE ROLL EXCITATIONS 0.8 ω'1 141.153 309.174 618.612 165.666 546.482 886.990
1.0 ω'1 716.198 903.424 1219.392 666.429 981.332 1409.812
Fig. 8 and Table 2 present the results for the sloshing
1.2 ω'1 183.924 321.814 465.267 87.447 236.756 347.834
pressure at point P1 in the model tank, obtained under pure
1.4 ω'1 79.544 160.115 277.848 13.921 79.483 159.855
sway and pure roll excitation forces of various frequencies
1.6 ω'1 40.974 123.624 254.184 19.642 116.830 228.883
and amplitudes. The data include the average values of the
top 10% sloshing pressures between the 16th and 21st cycles, 1.8 ω'1 109.614 184.812 276.799 73.855 191.390 303.609
after the transient phase.

moving it out-of-phase, as shown in Fig. 9. A substantial phase
shift occurred in Case 5 because the increased amplitude
of the roll excitation influenced the sloshing motions more
than the sway excitation. Case 5 exhibited the lowest sloshing
pressures due to significant force cancellation, as indicated
by the similarity between the average pressures for pure sway
with an amplitude of 0.015 m and pure roll with an amplitude
of 3° (see Table 2). Finally, Case 6 demonstrates that with
a large roll amplitude of 5°, the roll excitation dominates
the sloshing motion to the extent that the phase of pressure
Fig. 8. Average pressures at P1 for the prismatic model tank vs. excitation variations shifts to the opposite to that of pure sway, and
frequency of pure sway and pure roll also increases the pressure beyond the cancellation effect
observed in Case 5.
Table 3 provides information on the maximum peak
pressure during the initial 25s and the average values for
COUPLED SWAY-ROLL EXCITATIONS the top 10% of sloshing pressures at point P1 between the 16th
and 21st cycles. Case 1 showed an increase of approximately
We also investigated several cases of coupled sway-roll 200 Pa in the maximum peak pressure compared to pure
excitations for the model tank, which were divided into sway; although Cases 2 and 3 exhibited significantly higher
two types, with the opposite-phase or the same-phase. The maximum peak pressures compared to that of pure sway,
opposite-phase cases were Case 1 (roll amplitude: 1°), Case 2 the rise in the average pressures was not as significant as
(roll amplitude: 3°), and Case 3 (roll amplitude: 5°), while the the increase in the maximum peak pressures. The dramatic
same-phase cases were Case 4 (roll amplitude: 1°), Case 5 (roll increase in the maximum peak pressure (Pmax) can be
amplitude: 3°), and Case 6 (roll amplitude: 5°). In all cases, the attributed to the instability of the pressure associated with
amplitude of the sway excitation was fixed at 0.015 m, while the roof impact. In contrast, the maximum peak pressures
the roll motion was varied, with amplitudes of 1°, 3°, and 5°. and average pressures in both Cases 4 and 5 were lower than
The frequency of the coupled excitations was set to 1.0 ω'1. for pure sway. Case 6 showed higher pressures than for pure
Fig. 9 shows the variations in sloshing pressure between sway, as the cancellation of sloshing-induced forces was
the 16th and 21st cycles for Cases 1–6, together with the results insignificant due to the unbalanced forces of the roll with
for the pure sway scenario (amplitude: 0.015 m). It is seen that a large amplitude of 5°, which imposed a phase shift toward
in Cases 1–3 (coupled excitations with the opposite-phase), the opposite of the sway motion.
the sloshing pressures are in-phase, thus causing higher The movements of the free surface of the liquid inside
pressures compared to those for pure sway. Moreover, as the the tank under pure sway and in Cases 1–6 are presented in
amplitude of the roll excitation increases, sloshing pressure Fig. 10. It can be seen that under coupled sway-roll with the
also increases due to the intensified sloshing-induced forces opposite-phase, the free surface flows became more complex
under the coupled sway-roll excitations. The variation in with an increase in the amplitude of the roll. In contrast,
sloshing pressures becomes more unstable with an increase when the sway and roll had the same-phase, sloshing did not
in the roll amplitude, leading to complex pressure variations. exhibit significant oscillation compared to the opposite-phase
This phenomenon arises when the amplitude surpasses cases (Cases 1–3). As a result, Case 5 (Fig 10(f)), with a roll
a specific threshold, resulting in roof impact. As the amplitude amplitude of 3°, was the most stable, while Case 3 (Fig. 10(d))
increases, in Cases 2 and 3, the impact on the roof intensifies, exhibited the most unstable behaviour of the free surface.
leading to significant deviations from linear predictions and Tab. 3. Maximum peak pressures and average pressures of the prismatic model
the emergence of intricate pressure variations [26]. tank for each scenario
In Cases 4 and 5, it was observed that the sloshing pressures
Scenario Pmax [Pa] Time [s] Average pressure [Pa]
in the tank were decreased compared to those under pure
Pure sway 1,037.687 9.55 903.424
sway. This reduction pressure results from the cancellation of
sloshing-induced forces under coupled sway-roll excitations, Case 1 1,237.384 7.28 1,007.630
primarily due to the phase alignment of the tank’s excitation. Case 2 3,277.842 6.17 1,342.786
This phase alignment occurs in the direction where the Case 3 3,961.499 11.80 1,548.946
sloshing-induced forces generated in the tank by each sway Case 4 800.638 16.35 749.527
and roll cancel each other out. In addition, the magnitude of Case 5 568.884 18.01 537.824
the roll amplitude plays an important role in influencing this
Case 6 1,179.679 7.83 953.605
pressure decrease. In Case 4, with small roll amplitude of 1°,
the pressure variations are mostly in phase with pure sway,
but this results in a lower sloshing pressure compared to pure
sway. In Case 5, however, the increased roll amplitude causes
a deviation and shift in the phase of the pressure variation,

cases. To mitigate this noise,
a filtering technique based
on the FFT algorithm with
a cutoff frequency of 5 Hz was
employed [27].
In all cases, the analysis
revealed that the highest
amplitude of pressure occurred
at the dominant frequency
of the frequency domain,
0.88 Hz, which was closely
approximated to the primary
natural frequency of the
tank, 1.0 ω'1 (= 0.884918 Hz).
Other pressure peaks were
also observed at frequencies
corresponding to integer
multiples of the dominant
frequency, specifically at 1.76,
2.64, 3.52, and 4.40 Hz. This
observation suggests that in
addition to the dominant
frequency, these integer
multiples of the dominant
frequencies can also affect
Fig. 9. Comparison of pressure variations at P1 of the prismatic model tank sloshing behaviour significantly [28]. The amplitudes of
(over cycles 16 to 21): top graph (pure sway, Cases 1, 2, and 3) and bottom pressure at these frequencies are presented in Table 4, and
graph (pure sway, Cases 4, 5, and 6) pressure histories and FFT spectra are shown in Fig. 11.
In Case 1, the pressure amplitude at the dominant frequency
was larger than that for pure sway, whereas in Cases 2 and 3,
the amplitudes were very
similar to that for pure sway.
However, the amplitudes at
frequencies of 1.76 and 2.64 Hz
increased compared to those
of both pure sway and Case 1.
In Case 3, the amplitudes were
larger in the vicinity of 3.52 and
4.40 Hz (i.e., 3.48, 3.56, 4.24,
4.36, 4.44, and 4.52 Hz), rather
than at the 3.52 and 4.40 Hz.
Furthermore, in Cases 2 and 3,
Fig. 10. Contours of the volume fraction of the prismatic model tank: (a) pure sawtooth-shaped amplitude spectra were observed, implying
sway, (b) Case 1, (c) Case 2, (d) Case 3, (e) Case 4, (f) Case 5, (g) Case 6 the occurrence of complex sloshing phenomena. These
phenomena were attributed to the nonlinear characteristics
associated with irregular wave crests and breaking waves [29].
FFT ANALYSIS Fig. 11(c) and (d) demonstrate the instability of the sloshing
pressure due to roof impacts, contributing to a significant
The sloshing pressure data obtained from 0–25 s were increase in the pressure amplitude at or near the integer
subjected to quantitative analysis using the FFT algorithm. multiples of the dominant frequency (i.e., 1.76, 2.64, 3.52,
This analysis focused on the time series data representing the and 4.40 Hz). In Cases 4–6, the pressure amplitudes at the
sloshing pressures in the model tank under three scenarios: dominant frequency were smaller than that for pure sway.
pure sway and coupled sway-roll excitations with both This result is attributed to the cancellation effect between
the same-phase and the opposite-phase. The results were sloshing-induced forces due to coupled sway-roll excitations
examined in both the time domain and frequency domain, as with the same-phase. In Case 5, the pressure amplitude at
shown in Fig. 11. Throughout the FFT analysis, we observed the dominant frequency was the smallest, as the largest
the presence of high-frequency noise exceeding 5 Hz in all cancellation occurred compared to the other cases.

These FFT results revealed several phenomena. Firstly, CONCLUSION
for coupled sway-roll with the opposite-phase without roof
impact, the amplitudes of pressure increased at the dominant In this study, we investigated the liquid sloshing loads
frequency compare to that of pure sway. Secondly, for coupled exerted on the inner walls of a KC-1 membrane LNG tank
sway-roll having the opposite-phase with roof impact, the under various excitation scenarios. These scenarios included
amplitudes of pressure resulted in tallest peaks that are similar single excitations (pure sway and pure roll) and coupled sway-
in magnitudes and comparable to that of pure sway at the roll excitations, both with the same-phase and the opposite-
dominant frequency; while at or near integer multiples of the phase, across different ranges of excitation amplitudes and
dominant frequency, most of the pressure peaks increased frequencies. The findings revealed that the intensity of liquid
compare to that of pure sway. Thirdly, when the excitations are sloshing loads significantly increases when the resonance
coupled with the same-phase, the amplitude at the dominant effect occurs. Moreover, we observed that under coupled
frequency decreased compare to the sway’s. Finally, it was sway-roll excitations with the same-phase, the sloshing loads
confirmed that the amplitude at the dominant frequency become weaker than that for the case of pure sway, except
decreased significantly when the sloshing-induced forces when the force of the roll is substantially larger than the force
of sway and roll, coupled in the same-phase, were similar. of the sway, such that it overcomes the diminishing effect.
Conversely, when applied
with the opposite-phase,
coupled sway-roll excitations
strengthen and increase the
sloshing loads substantially
more than in the case of pure
sway.
FFT analysis further
identified that when the
excitation frequency was
1.0 ω'1, representing the first
natural frequency of the tank,
the sloshing pressure peaked
predominantly at the dominant
frequency, at 0.88 Hz, which
is close to ω'1 (= 0.884918 Hz).
Interestingly, under coupled
sway-roll excitations with the
same-phase, the amplitude
of sloshing pressure at
the dominant frequency
was significantly reduced
compared to that of pure sway,
contrasting with the cases with
Fig. 11. Pressure histories (with Pmax) and FFT spectra at P1 of the prismatic
model tank: pure sway - (a) and (a´); Case 1 - (b) and (b´); Case 2 - (c ) the opposite-phase, where it was increased. When coupled
and (c´); Case 3 - (d) and (d´); Case 4 - (e) and (e´); Case 5 - (f) and (f´); sway-roll excitations were in the opposite-phase with no roof
and Case 6 - (g) and (g´) impact conditions, the amplitude at the dominant frequency
Tab. 4. Amplitudes of sloshing pressure signals in the prismatic model tank
increased. However, in the presence of roof impact during
at frequencies under different scenarios sloshing, instabilities in the free surface manifest, resulting
in a nearly constant amplitude at the dominant frequency in
Frequency [Hz]
Scenario comparison to the sway’s. Concurrently, the amplitudes of
0.88 1.76 2.64 3.52 4.40 sloshing at or in proximity to other integer multiples of the
Pure sway 435.933 99.297 110.365 86.362 22.177 dominant frequency exhibit predominant increase.
Case 1 447.610 83.227 156.256 95.055 31.253 This study has demonstrated that a resonance effect in the
Case 2 433.445 140.796 216.485 100.433 52.718 context of the sloshing phenomenon in a KC-1 tank occurs
Case 3 437.104 212.115 179.821 50.733 29.098
when the natural frequency of the fluid sloshing inside
the tank aligns with the frequency of the external forces
Case 4 368.064 124.872 43.728 59.307 7.711
acting on the tank. This alignment can lead to significant
Case 5 216.485 98.965 8.619 21.374 6.767
amplification of the sloshing motion, potentially resulting
Case 6 411.733 133.329 126.097 79.448 18.885 in instability, the risk of fluid spillage, or even structural
damage. In addition, when the overall time series data for
sloshing pressure were analysed, instances of roof impact were

n.which This represents method provides the volume a solution occupied by the liquid
for effectively in each
tracking thecomputational
nonlinear freecell as a
surface
n. This method provides
s introduction, the VOF amethod solutionhas forbeen effectivelywidelytracking used tothe track nonlinear
the freefree surface
surface in
s introduction,
es, and it is alsothe adoptedVOF method in the present has been study, widely usingused Ansys to trackFluentthe free surface
software. Eq. (1) in
es, that
on and expresses it is also adopted the volume in the presentofstudy,
fraction liquid using occupying Ansysa Fluent cell. software. Eq. (1)
on that expresses the volume fraction of 0liquid occupying a cell.
ccupied by the liquid in each 𝛼𝛼(𝑥𝑥,computational
𝑦𝑦, 𝑧𝑧) = {0 ~ 0 1 cell as a (1)
dion byfor theeffectively
liquid in each tracking computational 𝑥𝑥, 𝑦𝑦,nonlinear
𝛼𝛼(the 𝑧𝑧) = {cell 0~ 1free
1as asurface (1)
found to introduce instability into the pressure variations,
1a surface in cell as a
ω1 First natural frequency of the rectangular tank [rad s-1]
od
volume the hasofliquid
effectively αbeenoccupied intracking
each
widely bycomputational
theliquid
used
the nonlinear
to track in cell
freeas
the
each surface
free
computational
ue is zero, making pressure predictions challenging. When the KC-1
this represents an empty state with no liquid in the cell, while values ω'1 First natural frequency of the prismatic tank [rad s-1]
es ctively
been
iedpresent
aand byα
solution tracking
widely isstudy,
the for
liquid used the
using
effectively
in nonlinear
torepresents
each track
Ansys the
tracking
computational free
Fluent free surface
surface
software.
the nonlinear
cell in
as Eq.a (1)
free surface
ue
ro of one zero,
represent tank is subjected to wave conditions involving both sway
this a free surface an empty
that state
includes with bothno liquid
liquid in
and the
gas.cell,
When while
thevalues
value P1 Internal pressure point in the rectangular tank and
OFnt
the
or widely
action study,
method
liquid
effectively of used
using
liquid
in haseachto Ansys
been
tracking track
occupying a Fluent
widely
computational
the free
used
isnonlinear surface
software.
apresent
cell. tothat
cell track
freeasthein
Eq. (1) free
the
a surface surface in
roindicatesand one represent
that and liquid
only roll free excitation
surface forces,
in includes
cell.there both
Fig. are measures
1 liquid
shows and
the gas. that
volume Whencan be value
the
fractions of prismatic tank [Pa]
udy,
fed of
ctively
sindicates
beenin using
liquid the tracking
widely Ansys
occupying
present the
used Fluent
study, atocell.
nonlinear software.
using
track free
theAnsys 𝛼𝛼.Eq.
free inFluent
surface (1) cell.
surface software.
inFig. the Eq. (1)the volume fractions of
the cells 0 that
based implemented
only on liquid
the values to
is present ofeffectively the mitigate 1 shows risk of substantial and Pmax Maximum peak pressure [Pa]
𝑦𝑦,
quid
folume
sent 𝑧𝑧0) study,
widely =occupying
{fraction
0used
~based 1 toof
using atrack cell.
liquid
Ansys thethe occupying
Fluent free of surface
𝛼𝛼.
software. a cell. in Eq. (1) (1)
the cells
1Ansys
unstableon pressure
values surges. These measures include dampening
{n0 of
dy, ~using 1liquid occupying Fluent a software.
cell. Eq. (1) (1)
1𝛼𝛼(
or 0controlling the sloshing motion using baffles or dampers,
0𝑥𝑥, 𝑦𝑦,state 𝑧𝑧) = with
1uidempty occupying {a0cell. ~ 1no liquid (1)while values (1)sloshing-induced
an promoting cancellation
in the cell, effects between the
= 1
pty {
ace0state ~ 1 with
that includes noforces liquidbothstemming in the and
liquid cell,from while
gas. each
When values excitation,
the value or adjusting the tank’s
(1) REFERENCES
1in the (1)
at ateincludes
esent
presents with noboth
an liquid
cell.
empty natural
liquid
Fig. in the
state 1and frequency.
showscell,
with gas.no while
the When values
volume
liquid the
in the value
fractions
cell, while of values
nof
mptyludes
the
free 𝛼𝛼. surface
cell.
statebothFig. liquid
with that 1no shows Further
and
includes
liquid gas.the inboth investigation
When
volume
the thefractions
liquid
cell, value
whileand gas. will
valuesofWhen be necessary
the value to explore the 1. H. N. Abramson, “The dynamic behavior of liquids in
cell.
uid te with
hat is Fig. no1liquid
present
includes shows insloshing-induced
both the in the
liquidthevolume
cell. cell,
Fig.gas.
and 1fractions
while showsforces
Whenvalues theof
theinvolume
more diverse
value fractions scenarios,
of including moving containers, with applications to space vehicle
udesvalues
in theboth of liquid
cell. 𝛼𝛼.
Fig.various 1and shows gas. types When
the of multiple
volume the value fractions coupled of excitations by manipulating technology,” Department of Mechanical Sciences, Southwest
cell. Fig. 1 shows thethe phases
volume of fractions
the excitations of with different filling heights. Research Institute, Washington, D.C., NASA-SP-106, 1966.
Moreover, it is important
to expand the analysis domain from
two dimensions to three, for a comprehensive understanding
2. L. Hou, F. Li, and C. Wu, “A numerical study of liquid
EQUATIONS of the sloshing phenomenon. sloshing in a two-dimensional tank under external
EQUATIONS excitations,” Journal of Marine Science and Application,
vol. 11, pp. 305-310, September 2012, doi: 10.1007/
y, a conventional numerical approach is employed to calculate the sloshing
ACKNOWLEDGEMENT s11804-012-1137-y.
y,liquid in a tank. It isnumerical
a conventional assumed that the fluidisflow
approach is incompressible,
employed to calculate and thethe turbulent
sloshing
hingliquidphenomenon
distribution inofa𝛼𝛼tank. values in Itcells represented
is assumed
[8] thatby thethe fluidReynolds-averaged
flow is incompressible, Navier-Stokes (RANS)
and the turbulent
Reynolds
hing
on of 𝛼𝛼 valuesphenomenon stresses
in cells [8] and
This the research
is represented turbulent bywas thesupported
flow field in thebyRANS
Reynolds-averaged the researchare fundbased
calculated
Navier-Stokes of on
(RANS) 3. B. F. Chen and C. H. Wu, “Effects of excitation angle and
model. The [8] volume
Dankook fraction University in 2021. on coupled heave-surge-sway motion on fluid sloshing in
Reynolds
values
ample ofinthe cells stresses
distribution ofand𝛼𝛼 valuestheinturbulent cells for[8] the flow twofield phases, in the liquidRANS and are gas,calculated
is calculated using
based
tionmodel. equation The volume
ution of 𝛼𝛼 values in cells [8]
in Eq. (2): fraction for the two phases, liquid and gas, is calculated using a three-dimensional tank,” Journal of Marine Science and
tion equation in Eq. (2): NOMENCLATURE Technology, vol. 16, pp. 22-50, December 2011, doi: 10.1007/
approach in1cells𝜕𝜕[8]is employed to calculate the sloshing
alues
[ (𝛼𝛼𝑞𝑞 𝜌𝜌𝑞𝑞 ) + 𝛻𝛻 ∙ (𝛼𝛼𝑞𝑞 𝜌𝜌𝑞𝑞 𝑢𝑢 ⃗ 𝑞𝑞 ) = 𝑆𝑆𝛼𝛼𝑞𝑞 + ∑𝑛𝑛𝑝𝑝=1(𝑚𝑚̇𝑝𝑝𝑝𝑝 − 𝑚𝑚̇𝑞𝑞𝑞𝑞 )] (2) s00773-010-0111-0.
hat
achthe 𝜌𝜌is
1𝑞𝑞 fluid𝜕𝜕𝜕𝜕
𝜕𝜕
[ employed (𝛼𝛼flow
𝑞𝑞 𝜌𝜌Denomination
is toincompressible,
𝑞𝑞 ) + 𝛻𝛻 ∙ (𝛼𝛼𝑞𝑞 𝜌𝜌𝑞𝑞 𝑢𝑢
calculate⃗ )the
[units]
and the turbulent
sloshing 𝑛𝑛
𝑞𝑞 = 𝑆𝑆𝛼𝛼𝑞𝑞 + ∑𝑝𝑝=1(𝑚𝑚̇𝑝𝑝𝑝𝑝 − 𝑚𝑚̇𝑞𝑞𝑞𝑞 )] (2)
ed isfluidby 𝜌𝜌 the
flow
𝑞𝑞
employed 𝜕𝜕𝜕𝜕 Reynolds-averaged
is incompressible,
to calculate Navier-Stokes
and
the the
sloshing turbulent (RANS)
merical approach
bles 𝑚𝑚̇𝑝𝑝𝑝𝑝 and 𝑚𝑚α̇ 𝑞𝑞𝑞𝑞 represent is employed to calculate thephasessloshing 𝑝𝑝 to 𝑞𝑞 and from phases 4. 𝑞𝑞 W. Wu, C. Zhen, J. Lu, J. Tu, J. Zhang, Y. Yang, K. Zhu, and
helent flow
Reynolds-averaged field in theflowVolume
RANS
Navier-Stokes
the fraction
mass transfer
are turbulent
calculated of
(RANS) liquid
based
from [-]onturbulent
dssumed
roach flow 𝑚𝑚 ̇ isisthat
incompressible,
the
employed 𝑚𝑚 ̇ fluid to and
calculate the
is incompressible, the sloshing and the 𝑝𝑝 𝑞𝑞 𝑞𝑞 J. Duan, “Experimental study on characteristic of sloshing
bles
or
ow y [9,
the field
eynolds-averaged
10].𝑝𝑝𝑝𝑝
two and𝑆𝑆
inphases,𝛼𝛼
the is
𝑞𝑞 the
α the
𝑞𝑞𝑞𝑞
RANS Volume
represent
source
liquid
q Reynolds-averaged areand
Navier-Stokes
the
term fraction
gas,
calculated
mass
for phase
isthe
(RANS) of
transfer
calculated
based phase
𝑞𝑞 in from
a q [-]
cell
on using (RANS)
phases
[9, 10]. to
Eq. and
(3) is from
the phases
governing
epresented
he sy [9, fluid
employed flow by isto incompressible,
calculate the andsloshing Navier-Stokes
turbulent
wo
eldmomentum, 10].
phases,
in 𝑆𝑆 is
𝛼𝛼liquid ρ the
as follows: Density
source
andcalculated
gas, term
isthe of the
for
calculated mixture
phase 𝑞𝑞
using in [kg/m
a cell ]
3[9, 10]. Eq. (3) is the governing
impact load in elastic tank with low and partial filling under
he theturbulent
flow isthe
Reynolds-averaged RANS
incompressible,𝑞𝑞flow are
field in the
Navier-Stokes
and RANS basedare
turbulent on calculated
(RANS) based on
phases,momentum,
raction liquid
for thethe as
andρ
two follows:
q RANS gas, Density
phases, isare calculated
liquid of phase
andusing q
gas, [kg/m 3
is calculated ] using rolling coupled pitching,” International Journal of Naval
flow
ynolds-averaged field in Navier-Stokes calculated (RANS) based on
⃗):𝑞𝑞 ) = 𝑆𝑆𝛼𝛼𝑞𝑞 𝜕𝜕+(𝜌𝜌𝑢𝑢
𝑢𝑢 ∑𝑛𝑛𝑝𝑝=1⃗)+ (𝑚𝑚𝛻𝛻Velocity
̇ 𝑝𝑝𝑝𝑝∙ (𝜌𝜌𝑢𝑢
−⃗𝑚𝑚𝑢𝑢 ⃗̇ 𝑞𝑞𝑞𝑞 )] −𝛻𝛻𝛻𝛻[m/s]
)vector
= + 𝛻𝛻 ∙ (𝜏𝜏𝜏𝜏) + (2)𝜌𝜌𝑔𝑔 + 𝐹𝐹 (3) Architecture and Ocean Engineering, vol. 12, pp. 178-183,
e two
eld in the phases, 𝜕𝜕𝜕𝜕
𝜕𝜕 liquid
RANS are and calculated gas, is calculated based on using
𝑆𝑆𝛼𝛼𝑞𝑞 + ∑𝑛𝑛𝑝𝑝=1 (𝑚𝑚
(𝜌𝜌𝑢𝑢̇ ⃗𝑝𝑝𝑝𝑝)q − + 𝑚𝑚 ̇ 𝑞𝑞𝑞𝑞
∙ (𝜌𝜌𝑢𝑢
𝛻𝛻Velocity )]⃗ 𝑢𝑢 ⃗ )vector= −𝛻𝛻𝛻𝛻 of+phase𝛻𝛻(2) ∙ (𝜏𝜏𝜏𝜏q) + [m/s]
𝜌𝜌𝑔𝑔 + 𝐹𝐹 (3) 2020, doi: 10.1016/j.ijnaoe.2019.10.003.
hases, liquid 𝜕𝜕𝜕𝜕 and gas, is calculated using
e+s∙mass ∑𝑛𝑛𝑝𝑝=1
(𝛼𝛼
the 𝜌𝜌 (𝑚𝑚
𝑢𝑢
velocity
𝑞𝑞 𝑞𝑞 𝑞𝑞 ⃗
transfer ̇ )
𝑝𝑝𝑝𝑝 = − 𝑚𝑚
𝑆𝑆
vector,
from 𝛼𝛼𝑞𝑞 ̇ +
𝑞𝑞𝑞𝑞 ) ]∑ 𝜌𝜌
Source
phases 𝑛𝑛the 𝑝𝑝density
𝑝𝑝=1 (𝑚𝑚 term
̇ to
𝑝𝑝𝑝𝑝 𝑞𝑞
− of𝑚𝑚
and oḟ phase
(2) )
the
from
𝑞𝑞𝑞𝑞 ] q in
mixture,
phases a 𝑞𝑞
cell 𝑝𝑝 [kg/m
the (2)static3
s] pressure, and 𝜏𝜏𝜏𝜏 the
m =s1, the
𝑆𝑆12].
transfer
for phase 𝜌𝜌𝑔𝑔
velocity
+ ∑
from 𝑛𝑛𝑞𝑞
and (𝑚𝑚
phases
in 𝐹𝐹
vector,
a ̇ cell − 𝑝𝑝
Mass
represent 𝜌𝜌
[9,𝑚𝑚 ̇
to the
10].𝑞𝑞 transfer
) ]density
and
theEq. from from
of
gravitational
(3) is thephase
phases
the 𝑞𝑞
mixture,
body
governing
(2) p to 𝑝𝑝phase
force the and q
static [kg/m
external s]
pressure,
3 body 𝜏𝜏𝜏𝜏 the
andforce, 5. B. Godderidge, S. R. Turnock, and M. Tan, “Evaluation of
𝛼𝛼𝑞𝑞 𝑝𝑝=1 𝑝𝑝𝑝𝑝 𝑞𝑞𝑞𝑞
+sfer
1,,∑
esent
hase 12]. 𝑛𝑛from
12].
𝑝𝑝=1 𝑞𝑞 (𝑚𝑚
The
the
in𝜌𝜌𝑔𝑔̇ phases
amass
𝑝𝑝𝑝𝑝 and−
stress
cell 𝑚𝑚 𝐹𝐹 𝑝𝑝
̇
transfer
[9, to
tensor,
𝑞𝑞𝑞𝑞 )
10]. ] 𝑞𝑞
Mass
represent and𝜏𝜏𝜏𝜏
from
Eq. , transfer
from
can
(3)the be
phases
is phasesfrom
𝑝𝑝
gravitational
theexpressed
(2) to
governing 𝑞𝑞𝑞𝑞phase
and as: body q
from to phase
force
phases andp
𝑞𝑞 [kg/m
external 3
s] body force, a rapid method for the simulation of sloshing in rectangular
ss𝑞𝑞, 12].
urce in
transferaterm Thefor
cell stress
[9,
from 10].
phasepphases tensor,
Eq.𝑞𝑞Static
(3)
in 𝑝𝑝a𝜏𝜏𝜏𝜏to
,is
cellpressure
𝑞𝑞
can
theand be
[9,governing
from
10]. [N/m
expressed Eq.phases
(3) ]
2 as:
is 𝑞𝑞the governing and octagonal containers at intermediate filling levels,”
𝜏𝜏𝜏𝜏 = 𝜇𝜇(𝛻𝛻𝑢𝑢 ⃗ + 𝛻𝛻𝑢𝑢 ⃗2𝑇𝑇 ) (4)
fer=
phase −𝛻𝛻𝛻𝛻
from +
phases 𝛻𝛻 ∙ (𝜏𝜏𝜏𝜏
𝑞𝑞 in a cell [9, 10]. Eq. 𝑝𝑝) +to Stress
𝑞𝑞
𝜌𝜌𝑔𝑔 and+ 𝐹𝐹 tensor
from phases[N/m 𝑞𝑞 ] (3) Computers & Fluids, vol. 57, pp. 1-24, March 2012, doi:
s: 𝜏𝜏𝜏𝜏 =(3) is⃗the
𝜇𝜇(𝛻𝛻𝑢𝑢 + governing
𝛻𝛻𝑢𝑢⃗ 𝑇𝑇 ) (4)
𝛻𝛻 + 𝛻𝛻 ∙ (𝜏𝜏𝜏𝜏 ) +
𝑞𝑞 in a cell [9, 10]. Eq. (3) is the governing
ynamic viscosity 𝜌𝜌𝑔𝑔 + [11,𝐹𝐹 Acceleration
12]. of gravity (3) [m/s 2
] 10.1016/j.compfluid.2011.09.010.
𝛻𝛻nsity ∙ (𝜏𝜏𝜏𝜏⃗ )𝑢𝑢
(𝜌𝜌𝑢𝑢
∙ynamic ⃗+
of 𝜌𝜌𝑔𝑔−𝛻𝛻𝛻𝛻
the
) viscosity
= + 𝐹𝐹 +[11,
mixture, ∙𝑝𝑝(𝜏𝜏𝜏𝜏
𝛻𝛻 External the
12]. ) +static 𝜌𝜌𝑔𝑔body 𝐹𝐹 force
+pressure, (3) [kg/m and 𝜏𝜏𝜏𝜏2the s2] (3)
e𝛻𝛻𝛻𝛻f gravitational
the 𝛻𝛻 ∙ (𝜏𝜏𝜏𝜏) +μ
+ mixture, 𝑝𝑝 𝜌𝜌𝑔𝑔
body the+Dynamic
static
𝐹𝐹
force and viscosity
pressure, external and body 𝜏𝜏𝜏𝜏 the
[Pa (3) s]force, 6. H. Jeong and W. Jaewoo Shim, “Calculation of boil-off gas
e𝜌𝜌 ∙mixture,
(𝜏𝜏𝜏𝜏
the
tational
be ) density
+ 𝜌𝜌𝑔𝑔 + 𝐹𝐹
expressed body 𝑝𝑝 the of
as: X
force static
the Horizontal
pressure,
mixture,
and external 𝑝𝑝 and
the 𝜏𝜏𝜏𝜏
displacement
static
body (3) the pressure,
force, [m] and 𝜏𝜏𝜏𝜏 the (BOG) generation of KC-1 membrane LNG tank with high
nal of body
ressed
esent thethe as:mixture,force
gravitationalX and 𝑝𝑝 the Amplitude
static
external body pressure,
body
force of the
force,
and sway
and 𝜏𝜏𝜏𝜏
external excitation
the body [m]
force, density rigid polyurethane foam by numerical analysis,”
(𝛻𝛻𝑢𝑢 ⃗ + 𝛻𝛻𝑢𝑢 ⃗ 𝑇𝑇 )the static 0
(4)
dmixture,
r,vitational , 𝑇𝑇can be𝑝𝑝body
𝜏𝜏𝜏𝜏as: expressed θ forceRotational pressure,
as:
and external body force, and 𝜏𝜏𝜏𝜏
displacementthe [°] Polish Maritime Research, vol. 24, no. 1, pp. 100-114, April
𝛻𝛻𝑢𝑢 ⃗ ) (4)
nal
xpressed body as:force θ and Amplitude
external body of the
force, roll excitation [°] 2017, doi: 10.1515/pomr-2017-0012.
) 𝜏𝜏𝜏𝜏 = 𝜇𝜇(𝛻𝛻𝑢𝑢 ⃗ +ω𝛻𝛻𝑢𝑢 0 𝑇𝑇
⃗ )Frequency [rad (4) (4)
d as: 𝑇𝑇 s -1
]
+ 𝛻𝛻𝑢𝑢 ⃗ ) t Time [s] (4) 7. C. W. Hirt and B. D. Nichols, “Volume of fluid (VOF)
12].
) H Total height [m] (4) method for the dynamics of free boundaries,” Journal of
H1 Height of the chamfered bottom corner [m] Computational Physics, vol. 39, no. 1, pp. 201-225, January
H2 Height excluding the corners [m] 1981, doi: 10.1016/0021-9991(81)90145-5.
H3 Height of the chamfered upper corner [m]
B Total breadth [m] 8. J. Haider, “Numerical modelling of evaporation
B1 Breadth of the chamfered bottom corner [m] and condensation phenomena,” Thesis, Institut für
B2 Breadth of the chamfered upper corner [m] Raumfahrtsysteme, Universität Stuttgart, Lampoldshausen,
θ1 Angle of the chamfered upper corner [°] 2013.
θ2 Angle of the chamfered bottom corner [°]
h Height of the water filling [m]

9. V. Singal, J. Bajaj, N. Awalgaonkar, and S. Tibdewal, “CFD 20. B. Godderidge, S. Turnock, C. Earl, and M. Tan, “The
analysis of a kerosene fuel tank to reduce liquid sloshing,” effect of fluid compressibility on the simulation of sloshing
Procedia Engineering, vol. 69, pp. 1365-1371, 2014, doi: impacts,” Ocean Engineering, vol. 36, no. 8, pp. 578-587,
10.1016/j.proeng.2014.03.130. June 2009, doi: 10.1016/j.oceaneng.2009.02.004.
10. A. B. Desamala, V. Vijayan, A. Dasari, A. K. Dasmahapatra, 21. C. Hu and M. M. Kamra, “An unstructured mesh method
and T. K. Mandal, “Prediction of oil-water flow patterns, for numerical simulation of violent sloshing flows,” Journal
radial distribution of volume fraction, pressure and velocity of Hydrodynamics, vol. 32, no. 2, pp. 259-266, April 2020,
during separated flows in horizontal pipe,” Journal of doi: 10.1007/s42241-020-0019-z.
Hydrodynamics, vol. 28, pp. 658-668, August 2016, doi:
10.1016/S1001-6058(16)60670-4. 22. B. Godderidge, S. Turnock, M. Tan, and C. Earl, “An
investigation of multiphase CFD modelling of a lateral
11. M. L. Hosain, U. Sand, R. Bel Fdhila, “Numerical sloshing tank,” Computers & Fluids, vol. 38, no. 2, pp. 183-
investigation of liquid sloshing in carrier ship fuel tanks,” 193, February 2009, doi: 10.1016/j.compfluid.2007.11.007.
IFAC-PapersOnLine, vol. 51, no. 2, pp. 583-588, 2018, doi:
10.1016/j.ifacol.2018.03.098. 23. H. Akyildiz and E. Ünal, “Experimental investigation of
pressure distribution on a rectangular tank due to the liquid
12. S. H. Rhee, “Unstructured grid based Reynolds-averaged sloshing,” Ocean Engineering, vol. 32, no. 11-12, pp. 1503-
Navier-Stokes method for liquid tank sloshing,” Journal of 1516, 2005, doi: 10.1016/j.oceaneng.2004.11.006.
Fluids Engineering, vol. 127, no. 3, pp. 572-582, 2005, doi:
10.1115/1.1906267. 24. J. H. Jung, H. S. Yoon, and C. Y. Lee, “Effect of natural
frequency modes on sloshing phenomenon in a rectangular
13. O. M. Faltinsen and A. N. Timokha, “An adaptive tank,” International Journal of Naval Architecture and
multimodal approach to nonlinear sloshing in a rectangular Ocean Engineering, vol. 7, no. 3, pp. 580-594, May 2015,
tank,” Journal of Fluid Mechanics, vol. 432, pp. 167-200, doi: 10.1515/ijnaoe-2015-0041.
April 2001, doi: 10.1017/S0022112000003311.
25. S. Wu and Y. Ju, “Numerical study of the boil-off gas
14. O. M. Faltinsen and A. N. Timokha, “Analytically (BOG) generation characteristics in a type C independent
approximate natural sloshing modes and frequencies in liquefied natural gas (LNG) tank under sloshing excitation,”
two-dimensional tanks,” European Journal of Mechanics Energy, vol. 223, no. 15, pp. 1-19, May 2021, doi: 10.1016/j.
– B/Fluids, vol. 47, pp. 176-187, September-October 2014, energy.2021.120001.
doi: 10.1016/j.euromechflu.2014.01.005.
26. K. P. Thiagarajan, D. Rakshit, and N. Repalle, “The
15. N. Parthasarathty, H. Kim, Y. H. Choi, and Y. W. Lee, air-water sloshing problem: Fundamental analysis and
“A numerical study on sloshing impact loads in prismatic parametric studies on excitation and fill levels,” Ocean
tanks under forced horizontal motion,” Journal of the Engineering, vol. 38, no. 2-3, pp. 498-508, 2011, doi:
Korean Society of Marine Engineering, vol. 41, no. 2, pp. 10.1016/j.oceaneng.2010.11.019.
150-155, 2017, doi: 10.5916/jkosme.2017.41.2.150.
27. H. Kim, P. Nanjundan, J. Jeon, and Y. W. Lee, “Numerical
16. M. Hinatsu, “Experiments of two-phase flows for the estimation on applying air-trapping mechanism to suppress
joint research,” In: Proceedings of the SRI-TUHH mini- sloshing loads in a prismatic tank,” Journal of Mechanical
Workshop on Numerical Simulation of Two-Phase Flows, Science and Technology, vol. 34, no. 7, pp. 2895-2902, 2020,
Ship Research Institute, Tokyo, Japan, pp. 12-19, 2001. doi: 10.1007/s12206-020-0621-6.
17. Y. H. Chen, Y. F. Yue, Y. Zhang, R. P. Li, and X. Xu, 28. X. Yuan, Y. Su, and P. Xie, “Frequency characteristics
“Numerical investigation of vibration suppression for the of sloshing resonance in a three-dimensional shallow-
combined device of non-Newtonian fluids coupled elastic water rectangular tank,” Journal of Marine Science and
baffle,” Journal of Applied Fluid Mechanics, vol. 16, no. 3, Engineering, vol. 10, no. 11, pp. 1792-1804, November 2022,
pp. 591-602, 2023, doi: 10.47176/jafm.16.03.1311. doi: 10.3390/jmse10111792.
18. H. Jin, Y. Liu, H. Li, and Q. Fu, “Numerical analysis of the 29. H. Kim, M. K. Dey, N. Oshima, and Y. W. Lee, “Numerical
flow field in a sloshing tank with a horizontal perforated study on sloshing characteristics with Reynolds number
plate,” Journal of Ocean University of China, vol. 16, no. 4, variation in a rectangular tank,” Computation, vol. 6, no. 4,
pp. 575-584, 2017, doi: 10.1007/s11802-017-3369-6. pp. 53-63, October 2018, doi: 10.3390/computation6040053.
19. O. Ubbink, “Numerical prediction of two fluid systems

with sharp interfaces,” Thesis, Department of Mechanical
Engineering, London University, January 1997.

10.2478/pomr-2023-0058
ANALYSIS OF THE ENVIRONMENTAL IMPACT OF THE HULL

CONSTRUCTION OF A SMALL VESSEL BASED ON LCA
Jacek Nakielski
Gdansk University of Technology, Institute of Naval Architecture, Poland
* Corresponding author: janak@pg.edu.pl (J.Nakielski)
Abstract
In recent years, issues related to the impact of human activity on the natural environment have become pressing, and the
challenge of global warming necessitates immediate action. To support environmental protection efforts, it has become
imperative to adopt a broader perspective when evaluating various products and systems. A valuable tool for such assessments
is a life cycle assessment (LCA), which enables a comprehensive analysis of the entire life cycle of a product.
This paper presents a comparative analysis of the hull of a fast patrol craft, fabricated using three different materials: steel,
aluminium, and composite materials. The LCA covers every stage from material production, through the construction
and use of the hull, to its eventual disposal. A specific criterion was established to evaluate the impact of the hull on the
environment, with clearly defined system boundaries.
In the final section, we draw some conclusions that underscore the importance of reusing construction materials.
By emphasising this approach, ecological footprints can be minimised and a sustainable future can be created.
Keywords: LCA, hull, craft, recycling materials
INTRODUCTION a comprehensive evaluation of the inputs and outputs of a product’s

environmental impact over its entire life cycle. An LCA typically
Over the past few decades, there has been a substantial has four main components: goal definition and boundary setting,
increase in awareness of ecological issues and the urgent need analysis, impact assessment, and interpretation [3].
for environmental protection. This heightened awareness has Due to the inherent complexity of a ship, conducting
primarily been driven by the alarming surge in greenhouse gas a thorough analysis is particularly challenging, and this especially
emissions, which significantly contribute to Earth’s changing true when interpreting the results. To streamline and enhance the
climate [1]. Numerous actions are now being carried out to effectiveness of the LCA method, it is advisable to break down
address this critical situation, including changes in approaches the ship into subsystems (hull, engine room, equipment, etc.).
to natural resources, production processes, and the exploitation, A crucial aspect of this process is the selection of appropriate
disposal and recycling of materials. system boundaries, as this can significantly influence the
To evaluate the environmental impact of a given product, the outcomes and their subsequent interpretation [4].
life cycle assessment (LCA) has emerged as a valuable method. In this paper, an analysis using LCA and life cycle cost analysis
This structured approach, which has been standardised by the (LCCA) is applied to a luxury mono-hull motor yacht [5]. Four
ISO (International Organization for Standardization) [26], enables design configurations of the hull are considered, and it is shown

that the lowest total life cycle cost is found for a configuration STUDY OBJECT
with an aluminium hull and carbon fibre composite hatches.
In a study by Wang et al. [6], an analysis based on an LCA The hull of the fast patrol craft described in [8] was chosen
was carried out for a hybrid ferry. LCA models were established as a subject for further analysis. The primary objective was to
using commercial software called GaBi (a portmanteau of two find the optimal shape of the hull for several different materials.
German words, Ganzheitliche Bilanzierung), and included Table 1 provides an overview of the fundamental parameters of
various activities associated with the four phases of the ship’s the unit, while Figs. 2 and 3 present visual representations of
life, including steel processing and machinery installations hulls made of aluminium and composite (fibre-epoxy laminate),
in the shipyard; the operation of the engine and batteries on respectively.
board; maintenance of the ship hull; and the scrapping of hull For this study, three materials were considered for the hull:
materials and machinery. Among other things, these authors steel, aluminium, and glass-epoxy laminate. Of the characteristic
presented an optimal strategy for coating of the hull. parameters examined here, one of the key factors was the weight
For a better understanding of the LCA methodology, the reader of the hull for each material, as shown in Table 2.
is referred to [7]. This paper presents several examples of decisions
Tab. 1. Specifications of the 24 m high-speed patrol
made based on the LCA methodology; for example, the decision craft studied in this paper [8]
as to whether the hull of the new Greenpeace flagship “Rainbow
Warrior III” was to be built from steel or aluminium was made by Basic parameter Value Units
the Dutch company TNO.35 using the LCA methodology. Based
on the results from the LCA, steel was chosen for the hull, while Overall length 23.85 m
the superstructure and mast were built from aluminium. Length between perpendiculars 20.05 m
Overall breadth 5.10 m

METHOD
Draft design waterline 0.97 m
Loaded displacement 48.20 tons

ENVIRONMENTAL IMPACT ASSESSMENT
CATEGORY FOR THE HULL Light displacement 38.20 tons
Total power installed 1380 kW

When assessing the environmental impact of a particular
object, the first step is to determine the factors that will define this
impact. Today, one of the most widely used criterion for evaluation Table 2 shows the weights of the hull when made of different
is the carbon footprint, which represents the total greenhouse gas materials.
emissions resulting from the production, operation, and disposal
Tab. 2. Mass of the hull of the 24 m high-speed patrol craft [8]
of the hull of a craft. Alternatively, the water footprint or the
overall environmental footprint can be considered. In this article, Material Mass [kg]
however, energy consumption serves as the chosen category for
evaluating the object’s environmental impact. Steel (NV40) 16,700
Data on the energy consumption of various technological
processes can be found in a diverse range of literature Aluminium (NV-5083) 8,700
sources. Although the appropriateness of using energy as an
environmental impact criterion is subject to debate, we note that Composite (fibre-reinforced epoxy) 7,700
energy exhibits a lower variability compared to other physical
flows. This enhances its reliability as a measure for assessing
the environmental influence of an object.
Fig. 1. Phases of an LCA according to ISO 14040:2006 Environmental Fig. 2. Structure of the aluminium hull [8]
management—Life cycle assessment—Principles and framework [2]

Fig. 3. Structure of the composite hull [8] Fig. 4. Boundaries of the system [2]
BOUNDARIES OF THE SYSTEM RESULTS
To conduct a comprehensive analysis based on LCA, it is vital

to establish clear boundaries and definitions. In the analysis PRIMARY PRODUCTION OF MATERIALS
presented below, the boundaries have been precisely defined
as follows [9]: To determine the energy consumption over the entire life
1. A „cradle-to-grave” approach is applied that considers the cycle of the hull, the initial step is to calculate the energy
entire life cycle of the hull, encompassing the production consumed in the production of each individual material. This
of materials required for its construction, its operational analysis includes the primary production of the material, starting
phase (lasting 20 years), eventual disposal, and the from the extraction procedure, and all necessary technological
potential for material reuse. processes; the energy needed to obtain materials from recycling
2. Characteristics of the hull that remain constant and do not is also considered.
vary depending on the material used have been excluded Table 3 presents a comprehensive summary of the results
from the analysis. for the energy consumption during production of the materials
3. The assumption is made that all materials and processes used to construct the hull of the craft. These findings serve
examined here are exclusively used to construct the hull, as a basis for a further evaluation of the hull’s environmental
and are not employed elsewhere. impact using the LCA approach.
4. Fuel consumption has been excluded from the analysis, given From the table above, it can be seen that the production of
that a lighter hull generally leads to reduced fuel consumption aluminium is the most energy-intensive process among the
or requires a less powerful engine for propulsion. materials considered here: for instance, the energy consumption
5. The production of tools and other materials essential for associated with the recycling of aluminium is comparable to
the construction of the hull is not considered in this study. the primary production of steel. However, for the construction
Clearly defining these boundaries allows the analysis to be of the same hull, approximately twice as much steel is needed
focused and relevant, and permits a more accurate evaluation of compared to aluminium. The production of composite materials
the hull’s environmental impact based on the selected criteria. is also a highly energy-intensive process.
The boundaries of the system are illustrated in Fig. 4.
Tab. 3. Total energy required to produce the hull material [4], [10], [11], [12]
Energy
Process Mass Units SEC Units Unit
consumption
Steel production (primary steel) 16,700 kg 22.00 MJ/kg 367,400 MJ
Steel production (recycled steel) 16,700 kg 8.60 MJ/kg 143,620 MJ
Aluminium production (primary aluminium) 8,700 kg 220.0 MJ/kg 1,914,000 MJ
Aluminium production (recycled aluminium) 8,700 kg 20.0 MJ/kg 174,000 MJ
Production of glass-epoxy laminate 7,700 kg 70.0 MJ/kg 539,000 MJ
Production of glass-epoxy laminate

Tab. 4. Energy consumed in the production of the steel hull
Energy
consumption
Cutting with an oxy-acetylene torch 262.00 m 0.25 MJ/m 65.50 MJ
MAG welding of transverse stiffeners 821.00 m 4.70 MJ/m 3,858.70 MJ
MAG welding of longitudinal stiffeners 821.00 m 5.50 MJ/m 4,515.50 MJ
MAG welding hull and deck plating 262.00 m 3.00 MJ/m 786.00 MJ
Painting – wetted surface area of the hull 148.00 m2 38.00 MJ/m2 5,624.00 MJ
Painting – unwetted surface area of the hull 77.84 m2 25.00 MJ/m2 1,946.00 MJ
Painting – deck area 99.74 m 2
25.00 MJ/m 2
2,493.50 MJ
Painting – surface of the inner part of the hull 508.71 m2 10.00 MJ/m2 5 087.6 MJ
Total: 24,493.50 MJ
SEC - specific energy consumption
Tab. 5. Energy consumed in the production of the aluminium hull

Energy
consumption
Water jet cutting 262.00 m 0.06 MJ/m 15.72 MJ
Friction stir welding – longitudinal stiffeners 630.00 m 1.20 MJ/m 756.00 MJ
Friction stir welding – transverse stiffeners 260.00 m 1.20 MJ/m 312.00 MJ
Friction stir welding – hull and deck plating 262.00 m 1.20 MJ/m 314.00 MJ
Painting – wetted surface area of the hull x 148.00 m2 28.00 MJ/m2 4,144.00 MJ
Painting – unwetted surface area of the hull x 77.84 m2 15.00 MJ/m2 1,167.60 MJ
Painting – deck area 99.74 m2 15.00 MJ/m2 1,496.10 MJ
Painting – surface of the inner part of the hull 554.95 m2 10.00 MJ/m2 5,549.52 MJ
Total: 13,755.34 MJ
Tab. 6. Energy consumed in the production of the composite hull

Energy
consumption
Plug 213.00 m2 51.00 MJ/m2 10,863.00 MJ
Steel supporting structure 213.00 m 2
10.00 MJ/m 2
2,130.00 MJ
Vacuum infusion process 213.00 m2 7.00 MJ/m2 1,491.00 MJ
Curing 213.00 m2 430.00 MJ/m2 91,590.00 MJ
Painting – wetted surface area of the hull 148.00 m 2
28.00 MJ/m2 4,144.00 MJ
Painting – unwetted surface area of the hull 77.84 m 2
15.00 MJ/m 2
1,167.60 MJ
Painting – deck area 99.74 m2 15.00 MJ/m2 1,496.10 MJ
Painting – surface of the inner part of the hull 485.58 m2 10.00 MJ/m2 5,855.83 MJ
Total: 117,737.53 MJ
HULL CONSTRUCTION Table 4 summarises the individual processes related to the

production of the steel hull.
To estimate the energy costs associated with the construction A very similar combination of individual postproduction
of the hull, including processes such as cutting, welding, vacuum processes are also found for the construction of the aluminium
infusion, hardening, and painting, specific parameters needed hull (see Table 5).
to be calculated, such as the weld lengths, mould surfaces, and The production technology used for the composite hull
surfaces to be painted. These calculations and estimates were differs significantly from the steel and aluminium hulls. An
made based on data from references [8] and [13]. important step in the production of the composite hull is the
We note that when constructing a hull using primary or implementation of a „plug” used as the basis for the lamination
recycled steel, the energy costs remain constant. Similarly, for process. For serial production, the plug can be reused, leading
an aluminium hull, the energy consumed during the production to reduced energy expenditure; however, for the purposes of
process is unchanged regardless of whether the aluminium used this article, unit production is assumed.
for production is from primary or recycled sources.

The results for the energy balance for the entire production Tab. 7. Energy consumed during use
process, including material production and hull construction, Energy
Process Units
are interesting. For primary steel, the hull construction process consumption
accounts for only 6.63% of the overall energy consumption at Primary steel
this stage, whereas for an aluminium hull, this value is only
0.72%. When the materials are obtained from recycling, these Total energy used to craft hull material 36,740.00 MJ
values rise to 39.09% for steel and 9.09% for aluminium. In
contrast, for a hull made using composite technology, the Total energy consumed in the hull fabrication 922.57 MJ
process
percentage is 21.84%. Fig. 5 shows a graph of these values.
Painting (every five years) 15,150.56 MJ
Total over 20 years 813,853.64 MJ
Secondary steel
Total energy used to craft hull material 14,362.00 MJ
Total energy consumed in the hull fabrication 922.57 MJ

process
Painting (every five years) 15,150.56 MJ
Total over 20 years 366,293.64 MJ
Primary aluminium

process
Painting (every year) 12,357.22 MJ
Total over 20 years 1,927,755.34 MJ
Secondary aluminium
Fig. 5. Total energy consumed in producing the hull material

and fabricating the hull
process
HULL USE
Painting (every year) 12,357.22 MJ
The next stage in the life cycle of the hull is its usage. In
the scenario considered here, we assume that each year, 10% Total over 20 years 597,940.64 MJ
of the material of the hull will be replaced, regardless of the
material or technology used for construction. Although other Composite material
scenarios could be considered, we adopt a simplified approach
for the sake of this analysis. In addition, we assume that the
life expectancy of each hull is 20 years, though this assumption Total energy consumed in the hull fabrication 10,607.40 MJ
may be subject to question, as hulls often remain in service process
beyond two decades. The replacement process itself, together Painting (every year) 11,663.53 MJ
with the energy consumed during the production of the
replacement material, makes up 10% of the energy required for Total over 20 years 1,501,692.60 MJ
the production of the original hull (in the case of the composite
material hulls, energy costs related to the production of the RECYCLING AND DISPOSAL
plug are not included).
Another factor related to hull usage is maintenance, which When considering the life cycle of the hull for a selected craft,
mainly involves painting. Steel hulls are repainted every five the final element is disposal, as steel and aluminium hulls can be
years, while those made of other materials require annual recycled. The disposal process involves cutting or dismembering
repainting. the hull, which naturally consumes energy. However, steel and
Table 7 provides a comprehensive overview of each type of aluminium offer extensive possibilities for reuse, with up to 95%
hulls and the individual energy costs during use. of the hull weight being recyclable according to [5]. In contrast,

Tab. 8. Recycling and disposal
Energy
consumption
Steel
Cutting with an oxy-acetylene torch 846.00 m 0.25 MJ/m 211.50 MJ
Reuse of steel −212 591.00 MJ
Total: −212 379.50 MJ
Aluminium
Plasma cutting 846.00 m 0.86 MJ/m 727.56 MJ
Reuse of aluminium −1 653 000.00 MJ
Total: −1 652 272.44 MJ
Composite material
Shredding 7,700.00 kg 0.92 MJ/kg 7,084.00 MJ
Burning 770.00 kg 30.00 MJ/kg −23,100.00 MJ
Total: −16,016.00 MJ
only 5% of a composite hull can be recycled [5]. Research is

under way to improve the recycling possibilities for composite
materials [14, 15].
DISCUSSION
Table 9 and Fig. 6 show the total energy consumption for

each type of material considered here. For recycled steel and
aluminium, the environmental benefits from their reuse were
not taken into account.
Tab. 9. Total energy consumption
Primary Recycled Primary Primary Composite
steel steel aluminium aluminium material
993.25 GJ 534.50 GJ 4,353.42 GJ 786.42 GJ 2,142.41 GJ
The data show that building a hull from primary aluminium

incurs the highest energy consumption, followed by a hull made
of composite materials. The use of recycled materials significantly
reduces the energy consumption, and this is particularly noticeable
for aluminium. Although steel hulls are the heaviest, steel is found
to be the most energy-efficient material. The energy cost for
Fig. 6. Total energy consumption
primary steel is only about 200 GJ higher than that of recycled
aluminium. The difference in energy cost between primary and
recycled aluminium is considerable, whereas the difference may be calculated with greater precision.
between virgin and recycled steel is relatively smaller. A hull made Technological processes such as welding or gas cutting, which
of recycled steel is 1.86 times less energy-consuming than a hull are based on electrical or chemical energy, are less energy-
made of virgin steel; in contrast, the ratio for aluminium is 5.34. efficient than processes based on mechanical energy, such as
Composite materials are relatively inefficient in terms of welding and water jet cutting.
energy consumption, as the recycling of these materials is The analysis of the hull production focused on the processes
challenging. Intensive research is ongoing to make the recycling themselves (cutting, welding, painting). However, the overall
of composite materials more efficient. energy required to make the hull is higher than that considered
here, because our analysis did not take into account factors
CONCLUSION such as transport, water consumption, electricity needed for
lighting and ventilation, etc. A comparison of the ratio of the
In this study, a relatively simple model was considered with energy used in the production of the hull to the energy needed
similar life cycle scenarios for hulls made of different materials. to produce the material (0.72% for aluminium, 6.63% for steel)
In future analyses, the life cycle scenarios for specific materials need not take this additional energy into account, but in future

research, the total energy cost of hull production, especially for Naval Architecture, Marine Structures, Volume 24(1): Pages
recycled and composite materials, could be calculated to give 43-59, 2010. doi: 0.1016/j.marstruc.2011.01.001
more accurate results.
The production stage of the individual materials incurred 9. P. Borowiec, ‘Life cycle assessment for selected vessel’, MSc
the highest energy cost, meaning that the development of [Dissertation], Gdansk University of Technology, 2022.
technologies to reduce energy consumption during production
processes is crucial. It is also important to ensure that the 10. S. V. Joshi, L. T. Drzal, A. K. Mohanty, and S. Arora, ‘Are
energy used in the production of steel or aluminium comes natural fiber composites environmentally superior to glass
from renewable sources. fiber reinforced composites?’, Composites Part A: Applied
The choice of hull material significantly influences both the Science and Manufacturing, Volume 35, Issue 3, Pages 371-
energy costs and the environmental footprint. Analyses such as 376, 2004. doi: 10.1016/j.compositesa.2003.09.016
the one presented here can aid in making informed decisions
at the design stage. 11. P. Michaelis, T. Jackson, and R. Clift, ‘Exergy analysis of the
One key element of a strategy for reducing the cost of energy life cycle of steel’, Energy Volume 23, Issue 3, Pages 213-220,
consumption, and hence the impact on the environment, is 1998. doi: 10.1016/S0360-5442(97)00081-9
the reuse of materials. In the case considered here, this was
particularly applicable to aluminium. 12. G. Rombach and D. Liebig, ‘A scenario to optimise the energy
demand of aluminium production depending on the recycling
quota’, Resources, Conservation and Recycling, Volume 33, Issue
REFERENCES 3, Pages 217-234, 2001. doi: 10.1016/S0921-3449(01)00086-6.
1. J. Barreiro, S. Zaragoza, and V. Diaz-Casas, ‘Review of ship 13. F. Nicolae, C. Popa, and H. Beizadea, ‘Applications of life cycle
energy efficiency’, Ocean Engineering, Volume 257, 2022. doi: assessment (LCA) in shipping industry’, 14th International
10.1016/j.oceaneng.2022.111594. Multidisciplinary Scientific GeoConference & EXPO SGEM2014,
Section Air Pollution and Climate Change, Albena, Bulgaria,
2. ISO 14040:2006 Environmental management - Life cycle 17-26 Jun 2014. doi: 10.5593/sgem2014/b42/s19.038.
assessment - Principles and framework.
14. Jianqiao Wu, Ying Pan, Ziyi Ruan, Ziji Zhao, Jing Ai, Jinghai
3. S.D. Chatzinikolaou and N.P. Ventikos, ‘Assessing environmental Ban, and Xianghai Jing, ‘Carbon fiber-reinforced epoxy with
impacts of ships from a life cycle perspective’, in 2nd International 100% fiber recycling by transesterification reactions’, Frontiers
Conference on Maritime Technology and Engineering, MARTECH in Materials Sec. Polymeric and Composite Materials, Volume
2014, Lisbon, Portugal, 15-17 October 2014. doi.org/10.1201/ 9, 2022. doi: 10.3389/fmats.2022.1045372.
b17494
15. M. M. Rabby, M. Rahman, P. P. Das, V. Vadlamudi and
4. L. Joosten, ‘The Industrial Metabolism of Plastics. Analysis R. Rassel, ‘Carbon fibers recycling from degraded prepress and
of Material Flows, Energy Consumption and CO2 Emissions mechanical properties of recycled composite’, North America
in the Lifecycle of Plastics’, PhD [Dissertation], Utrecht Society for the Advancement of Material and Process Engineering
University, 2001. (SAMPE), Seattle, United States, 17 April 2023. doi: 10.33599/
nasampe/s.23.0267.
5. C. Favi, M. Germani, F. Campi, M. Mandolini, S. Manieri,
M. Marconi, and A. Vita, ‘Life cycle model and metrics in
shipbuilding: How to use them in the preliminary design
phases’, in 25th CIRP Life Cycle Engineering (LCE) Conference,
Copenhagen, Denmark, 30 April – 2 May 2018. doi: 10.1016/j.
procir.2017.11.071
6. H. Wang, E. Oguz, B. Jeong, and P. Zhou, ‘Life cycle cost and

environmental impact analysis of ship hull maintenance strategies
for a short route hybrid ferry’, Ocean Engineering, Volume 161,
Pages 20-28, 2018. doi: 10.1016/j.oceaneng.2018.04.084
7. T. Pawlik, Ed., ‘Ship life cycle assessment and management’,

City University of Applied Sciences and GAUSS GmbH,
Bremen, Germany, 2011.
8. I. Stenius, A. Rosén, and J. Kuttenkeuler, ‘On structural design

of energy efficient small high-speed craft’, KTH Centre for

10.2478/pomr-2023-0059
EFFECTS OF PROPELLER FOULING ON THE HYDRODYNAMIC

PERFORMANCE OF A MARINE PROPELLER
Ali Zinati
Department of Maritime Engineering, Amirkabir University of Technology, Tehran, Islamic Republic of Iran
Mohammad Javad Ketabdari
Department of Maritime Engineering, Amirkabir University of Technology, Tehran, Islamic Republic of Iran
Hamid Zeraatgar *
Faculty of Marine Technology, Amirkabir University of Technology, Tehran, Islamic Republic of Iran
* Corresponding author: hamidz@aut.ac.ir (Hamid Zeraatgar)
Abstract
Propeller performance is typically considered under clean conditions, despite the fact that fouling is an inevitable
phenomenon for propellers. The main objective of this study is to investigate the effects of roughness due to fouling on
the performance of a propeller using a CFD simulation in conjunction with the roughness function model. A simulation
of a clean propeller is verified for a five-blade propeller model using existing experimental results. A roughness
function model is then suggested based on existing measured roughness data. The simulations are extended for the
same propeller under varying severities of roughness. Initially, it is concluded that KT and ηo gradually decrease with
increasing fouling roughness, while KQ increases, compared to smooth propeller. For instance, at J=1.2 for medium
calcareous fouling, KT is reduced by about 26%, KQ increases by about 7.0%, and ηo decreases by 30.9%. In addition, for
the rough propeller, the extra power required is defined as the specific sea margin (SSM) to compensate for the power
loss. A slight roughness causes a large decrease in ηo. A propeller painted with foul-release paint and an unpainted
propeller are found to require 2.7% SSM and 57.8% SSM over four years of service, respectively. Finally, the use of
foul-release paints for propeller painting is strongly advised.
Keywords: Propeller performances, Blade roughness, Frictional resistance, CFD simulation, Fouling
introduction [3-10]. Temporary roughness, which refers to temporal

changes in the hull and propeller surface roughness, is
Nowadays, shipping is a more important means of caused by fouling organisms during a period of service [8].
transportation compared to land and aviation. World The increase in skin resistance due to fouling is responsible
globalisation has increased shipping traffic, transportation, for a significant proportion of the total resistance, as a small
and the capacities of goods transporters, which has increased amount of fouling can cause a significant increase in fuel
fuel consumption as a result. Ships are a contributor to consumption and air pollution.
greenhouse gas (GHG) emissions [1], and in recent years, The effects of roughness on hulls and propellers can be
increasing pressure has been placed on the marine industry evaluated using the boundary layer similarity law or the CFD
to decrease GHG emissions through regulatory legislation. method. Both of these methods require a roughness function
IMO (International Maritime Organization) indicated for the surface in question.
that efficiency improvements could be achieved through As mentioned at ITTC 2021 [11], there is a need to adopt
operational [2] and technological methods, as these could or develop new methods to predict the roughness effects of
increase overall performance and decrease fuel consumption marine biofouling and modern fouling-control coatings on

ship hydrodynamics and propeller performance. For this as a means to diminish the roughness are also considered.
reason, ITTC 2021 recommended that the roughness function The novelty of this study is that it sheds light on the details of
should be determined or developed by researchers for different the propeller performance with respect to surface roughness
surface conditions [11]. and presents a new roughness function model.
Several roughness function models have been proposed
that are appropriate for different surface conditions. Some of EFFECTS OF ROUGHNESS ON THE BOUND
these consider antifouling coatings such as the Townsin [12], EFFECTS OF ROUGHNESS
Demirel [13], Vargas [14], and Grigson of Colebrook types ON THE BOUNDARY LAYER
[15]. One roughness function model proposed by Song [16] EFFECTS OF ROUGHNESS ON THE BOUND
was appropriate for polished surfaces with sandpaper A 60 and
turbulent A turbulent
boundary layer boundary
is assumed layer is to assumed
consist to ofconsist
two of two
regions: an i
80. Another roughness function model for surfacesThe covered
flow EFFECTSregions:
in the inner area an inner area
OFisROUGHNESS and an outer
affected by the surface area. The flow in
ON roughness, the
THE BOUND inner where
with closely packed sand grain roughness was proposed by area is affected by the surface roughness, whereas the flow
Cebeci [17]. The effects of biofilm can be predicted A
is turbulent
using
independent
the in
of the surface
boundary
the outer area layeris is conditions.
assumed of
independent tothe consist
surface ofconditions.
two regions: an i
roughness function model of Farkas [18]. However,The research
flow in the The velocity
inner areaprofile for smooth
is affected by walls
the surfacein the log-law region where
roughness,
The velocityboundary profile for smooth walls in the log-law region of a an
turbin
on roughness function models is still needed, as A indicated
turbulent of a turbulent
is independent of the surface conditions. boundary
layer is assumedlayer can to be defined
consist ofas follows:
two regions:
by ITTC 2021 [11] for the prediction of the roughness defined
Theeffects as follows:
flow in the inner area is affected by the surface roughness, wherea
of marine biofouling. The velocity profile
is independent of the for surface smooth walls
conditions. 1 in the log-law region of a turbu
The impacts of hard fouling and biofilmdefined on theas follows: 𝑈𝑈 + = 𝑙𝑙𝑙𝑙 (𝑦𝑦 + ) + 𝐵𝐵 (1)
performance of ships using the roughness functionThe proposed 𝜅𝜅
where velocity
y+ is the profile for smooth walls
non-dimensional normal in distance
the log-law fromregion the wall, of aκturbu
is th
by Grigson were studied by Farkas et al. [19, 20]. The impact +
1 +
defined as
B is the smooth follows: 𝑈𝑈 = 𝑙𝑙𝑙𝑙 (𝑦𝑦 ) + 𝐵𝐵
of inhomogeneous roughness distribution on the frictional wherewall y is log-law
+ intercept.
the non-dimensional 𝜅𝜅 normal distance from the
resistance of a plate was considered using the roughness where y+ is the κ is the von Karmannormal
wall,non-dimensional constant,
1 and B is the
distance from smooth
the wall,wall κ is th
function model in conjunction with a longitudinal roughnessRoughness log-law leads intercept.
to a decrease 𝑈𝑈 + = in 𝑙𝑙𝑙𝑙the (𝑦𝑦 +log-law
) + 𝐵𝐵 velocity profile, and
B is the smooth wall log-law intercept. 𝜅𝜅
position [21]. velocity
where y+ is theRoughness
profilenon-dimensional
for rough leadswallsto a decrease is
normal known indistance
the as log-law
thefrom velocitytheprofile,
roughness wall, function,
κ is the
The roughness effects of fouling conditions on
profile
Roughnessthe for and
rough
leads the downward
walls
to a in the
decrease shift of
turbulent
in the thevelocity
boundary
log-law profile for
layer
velocity rough
is walls
[25]:
profile, and
performance of some propellers have been investigated B isusing
the smooth is known wallaslog-law the roughness intercept. function, ΔU+. The log-law velocity
velocity profile for rough walls1 is known as the roughness function,
CFD simulations. Kellett et al. [22] investigated the effect of profile for rough walls in the turbulent boundary layer is [25]:
Roughness
profile leads to a decrease 𝑈𝑈 + = in 𝑙𝑙𝑙𝑙(𝑦𝑦the+)
+ 𝐵𝐵 − ∆𝑈𝑈 +velocity
log-law profile, and
biofouling on a real four-blade ship propeller at the modelfor rough walls in the turbulent 𝜅𝜅 boundary layer is [25]:
scale using the roughness function approach. Owen velocityet al. profile
Various parameters for rough can bewalls used1 to is define
knownroughness, as the roughness but the most function,
freq
+ +) +
[23] calculated the performance of a PPTC propeller profileunder for rough walls
height, ks. The dimensionless form in 𝑈𝑈
the = 𝑙𝑙𝑙𝑙(𝑦𝑦
turbulent + 𝐵𝐵
boundary − ∆𝑈𝑈 layer
of the roughness height is the roug is [25]: (2)
𝜅𝜅
different fouling conditions using a previously developed + ⁄
defined as
Various parameters 𝑘𝑘 𝑠𝑠 = 𝑘𝑘 𝑈𝑈
𝑠𝑠 can𝜏𝜏 𝜐𝜐 , where U
be used1 toτ define is the friction
roughness, velocitybut the. most frequ
roughness function. Song et al. [24, 25] investigated the 𝑈𝑈 + = 𝑙𝑙𝑙𝑙(𝑦𝑦 + ) + 𝐵𝐵 − ∆𝑈𝑈 +
roughness effect of barnacles with varying sizes andheight, coverage
The most
ks. TheVarious
common
dimensionless parameters
method of
form
𝜅𝜅can be
solving
of used
the roughness
for to the
define height isbut
roughness,
turbulent boundary
the rough
la
+ ⁄𝜐𝜐,be
of a KP505 propeller using a roughness function of the V arious
defined as 𝑘𝑘the
Grigsonparameters 𝑠𝑠 =most 𝑘𝑘𝑠𝑠 𝑈𝑈can
𝜏𝜏frequently
used
where usedUtoτ define
is the roughness,
the roughness
friction but the
height,
velocity .k s. Themost frequ
type. The impact of biofilm on propeller performance
implement
height, wasks. The
adimensionless
wall function
dimensionless approach
form of of using
the roughness
form an appropriate
height is the
the roughness roughness
roughness
height is the rough f
studied by Farkas et al. [26-28]. These studies havedefined
indicated
The most Reynolds
common
+ numberk sof
method +
, defined
solving as kfor +
= kthe U /v, where U boundary
turbulent is the lay
as 𝑘𝑘 𝑠𝑠 = 𝑘𝑘𝑠𝑠 𝑈𝑈𝜏𝜏 ⁄𝜐𝜐, where U τ is the sfriction s τ velocity.τ
that biofilms significantly decrease propeller performance, friction velocity.
implement a wall function approach using an appropriate roughness fu
and should therefore be given due importance. The most common
PROPULSION The CHARACTERISTICS
most method common method offor
of solving solving for the turbulent
the turbulent boundary lay
CFD-based hydrodynamic analysis has been extensively boundary layer flow near the wall is to implement a wall
implement a wall function approach using an appropriate roughness fu
employed in many areas of research. However, the literature function approach using an appropriate roughness function
review above indicates that CFD cannot represent thePROPULSION
complex model.CHARACTERISTICS
geometry of a rough surface such as a propeller. ITTC 2021
For a propeller in open-water conditions, the thrust coefficient, KT,
[11] recommended employing either the similarity law or PROPULSION CHARACTERISTICS
CFD simulations in conjunction with the roughness function
efficiency,0, areCHARACTERISTICS
PROPULSION expressed in non-dimensional forms. When the prop
model in order to include roughness effects. The For
themaina propeller
incoming Forinais
flow open-water
non-uniform,
propeller in conditions, and the conditions,
open-water the thrust the
quasi-propulsive coefficient,
thrust KT,
efficiency
objective of this study is to investigate the effects of roughness
efficiency,
of the open-water coefficient,
0 , are expressed K
efficiency,T
, torque inO, the relative
coefficient,
non-dimensional K Q
, and efficiency,
forms.
rotative efficiency, η
When 0
, aretheR, a
prop
on propeller performance through the use of CFD simulation
For
the a propeller
incoming
follows: expressed
flowin is in
open-waternon-dimensional
non-uniform, andforms.
conditions, theWhen thrust
the quasi-propulsive the propeller
coefficient, is KT,
efficiency
with a roughness function model. In addition, an attempt fitted aft
0, are expressed of the hull, the incoming flow is non-uniform, and
is made to introduce a new roughness function model.
efficiency,
of the To open-water efficiency,inefficiency
the quasi-propulsive
non-dimensional
O, the relative
𝜂𝜂𝐷𝐷 = 𝜂𝜂coefficient,
forms.
rotative When theprop
efficiency,
ηD, is a function of R, an
𝑂𝑂 𝜂𝜂𝐻𝐻 𝜂𝜂𝑅𝑅
the incoming flow is non-uniform,
follows: the open-water efficiency, ηO, the relative rotative efficiency,
achieve these aims, a five-blade propeller model is considered. and the quasi-propulsive efficiency c
The clean propeller is simulated and compared toofexisting the open-water ηR, and efficiency,
the hull efficiency, O, the ηH,relative
as follows: rotative efficiency, R, an
experimental results for validation, and the simulations are 𝜂𝜂𝐷𝐷 = 𝜂𝜂𝑂𝑂 𝜂𝜂𝐻𝐻 𝜂𝜂𝑅𝑅
follows: NUMERICAL SIMULATION
then extended to represent several severities of roughness
for the same propeller using the verified simulation setup. 𝜂𝜂𝐷𝐷 = 𝜂𝜂𝑂𝑂 𝜂𝜂𝐻𝐻 𝜂𝜂𝑅𝑅 (3)
The results are analysed, and the extra power required In thistosection, the governing equations are introduced, the roughnes
NUMERICAL SIMULATION
compensate for the power loss of the roughed propeller is
the numerical simulations of smooth and roughed propellers are expla
formulated and estimated. The effects of painting propellers
In this section, the governing equations are introduced, the roughnes NUMERICAL SIMULATION
the numerical simulations of smooth and roughed propellers are expla
62
In this section, the governing equations are introduced, the roughness
POLISH MARITIME RESEARCH, No 4/2023
the numerical simulations of smooth and roughed3 propellers are expla
3
NUMERICAL SIMULATION the impact of roughness on ΔU+ depends on the type and
coverage of the roughness.
In this section, the governing equations are introduced, In this study, the roughness parameters measured by
the roughness function is presented, and the numerical Schultz and Flack [29] are used to develop a new roughness
simulations of smooth and roughed propellers are explained. function model. The proposed model is fitted to the roughness
function values reported by Schultz and Flack [29]. One
MATHEMATICAL FORMULATION advantage of our roughness function model is that it can be
MATHEMATICAL FORMULATION applied to all types of fouled surfaces and typical antifouling
The governing equations in this context are those of mass coatings. The values of the sand grain roughness height and
MATHEMATICAL FORMULATION
and momentum conservation, which for compressible flows other relevant data are shown in Table 1 for a range of surface
in the Cartesian coordinate system are as follows [23]: conditions.
The governing equations in this context are those of mass and momentum conservation, which for
compressible
MATHEMATICAL the Cartesian coordinate system are as followsTab.
flows inFORMULATION
MATHEMATICAL
1. Representative coating and fouling conditions [23]
[23]:
The governing equationsFORMULATION
in this context are those of mass and momentum conservation, which for NSTM
CAL FORMULATION Description of condition ks (µm) Rt50 (µm)
𝜕𝜕(𝜌𝜌𝑢𝑢̅𝑖𝑖 ) coordinate system are as follows [23]:
compressible flows in the Cartesian rating*
(4)
(4) =0
The governing equations in𝜕𝜕(𝜌𝜌𝑢𝑢 𝜕𝜕𝜕𝜕context
𝑖𝑖 Hydraulically smoothwhich surfacefor 0. 0 0
𝜕𝜕(𝜌𝜌𝑢𝑢 ̅𝑖𝑖are) those 𝜕𝜕 inofthis
this ̅𝑖𝑖 ) are those of mass and momentum
𝜕𝜕𝜏𝜏̅̅̅
𝜕𝜕𝑝𝑝̅conservation,
conservation,
(4)
The
quations governing
in this context
compressible equations flows + in (𝜌𝜌𝑢𝑢̅𝑖𝑖 𝑢𝑢𝜕𝜕𝜕𝜕
the mass
Cartesian context =
̅̅̅̅̅̅
̅𝑗𝑗 + 𝜌𝜌𝑢𝑢𝑖𝑖 𝑢𝑢𝑗𝑗 ) = −
and ′ 0
momentum
coordinate ′ are thosesystem +
of are mass
𝑖𝑖𝑖𝑖
as and
which
follows momentum
for
[23]: conservation, which (5)
for
𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕𝑖𝑖system 𝑖𝑖 𝜕𝜕𝜕𝜕𝑖𝑖 system 𝜕𝜕𝜕𝜕𝑗𝑗 are asDescription Typical asofapplied condition AF coating 0
NSTM rating* 30 150 ks (µm) Rt50
compressible
ws in the Cartesian flows
𝜕𝜕(𝜌𝜌𝑢𝑢 ̅𝑖𝑖 ) in the
𝑖𝑖
coordinate
𝜕𝜕 Cartesian are
𝜕𝜕(𝜌𝜌𝑢𝑢̅𝑖𝑖 )
as followscoordinate [23]:
𝜕𝜕𝑝𝑝̅ 𝜕𝜕𝜏𝜏̅̅̅ follows [23]:
𝑖𝑖𝑖𝑖 HydraulicallyDeteriorated coating or light slime 10–20 (5) 100 300
where  𝜕𝜕(𝜌𝜌𝑢𝑢 is ̅the density, + 𝜌𝜌𝑢𝑢
(𝜌𝜌𝑢𝑢 ̅̅̅̅̅̅
′
̅𝑖𝑖𝑖𝑖𝑢𝑢 ′
𝑢𝑢̅𝑗𝑗𝑗𝑗𝜕𝜕𝜕𝜕+
is𝜌𝜌𝑢𝑢the ̅̅̅̅̅̅′ ′
=𝑖𝑖0𝑢𝑢𝑗𝑗 )Reynolds =− + stress, 𝑢𝑢 ̅ is (4) the
smooth surface
averaged (4)
Cartesian component
0. 0
𝜕𝜕𝜕𝜕𝑖𝑖 𝜕𝜕𝜕𝜕𝑗𝑗 Typical𝑖𝑖 as applied
(5)
𝑖𝑖 ) 𝜕𝜕𝜕𝜕
= 0𝑖𝑖𝜕𝜕(𝜌𝜌𝑢𝑢̅ )𝜕𝜕𝜕𝜕𝑖𝑖 𝜕𝜕 𝜕𝜕(𝜌𝜌𝑢𝑢 Heavy AF coating
slime 30 0 300 600 30 1
̅𝑖𝑖𝑖𝑖 ) 𝜕𝜕𝑝𝑝̅ 𝜕𝜕𝜏𝜏
̅̅̅ (4)
velocity, and 𝜕𝜕𝜕𝜕𝑖𝑖 p is the mean 𝑖𝑖
̅̅̅̅̅̅
pressure. = 𝜏𝜏
̅̅̅
0 are the 𝑖𝑖𝑖𝑖
mean Deteriorated
viscous coating
stress or
tensorlight slime
components, (5) as follows: 10–20 100 3
𝜕𝜕  is the ′ density,
+ 𝜌𝜌𝑢𝑢 ′ 𝑢𝑢 𝑖𝑖′𝑢𝑢
̅̅̅̅̅̅ Small calcareous fouling or weed 40–60 1,000 1,000
where
𝑢𝑢̅𝑖𝑖 ) 𝜕𝜕𝜕𝜕 𝑖𝑖 𝜕𝜕𝑝𝑝̅ 𝜕𝜕𝜕𝜕 𝑖𝑖
(𝜌𝜌𝑢𝑢
𝜕𝜕𝜏𝜏
̅̅̅
𝑖𝑖
𝑖𝑖𝑖𝑖
̅𝜕𝜕𝜕𝜕
𝑗𝑗
̅𝑗𝑗is + 𝜌𝜌𝑢𝑢
𝑖𝑖 the ′ ′
𝑖𝑖 𝑢𝑢𝑗𝑗 ) =−
𝑖𝑖𝑖𝑖Reynolds
𝜕𝜕𝜕𝜕 𝑖𝑖
+
𝜕𝜕𝜕𝜕 𝑗𝑗
stress, Heavy 𝑢𝑢
̅ 𝑖𝑖 is
slime the
(5) averaged Cartesian component 30 300 6
+ (𝜌𝜌𝑢𝑢̅𝑖𝑖 𝑢𝑢̅𝑗𝑗 + 𝜕𝜕(𝜌𝜌𝑢𝑢
𝜌𝜌𝑢𝑢̅̅̅̅̅̅ ′) = −
𝑖𝑖 𝑢𝑢𝑗𝑗̅ Medium calcareous
or weed fouling 70–8040–60 3,000 3,000 1,000
where  ρ is the ) density, 𝜕𝜕𝑖𝑖 + 𝜕𝜕𝜕𝜕 𝜌𝜌𝑢𝑢̅̅̅̅̅̅
′ 𝑢𝑢 ′ is is the Reynolds 𝜕𝜕𝑝𝑝̅stress, 𝜕𝜕𝜏𝜏
̅̅̅ Small calcareous
is the
fouling 1,0
𝑖𝑖 𝑖𝑖′̅̅̅
𝜏𝜏𝑢𝑢𝑖𝑖𝑖𝑖
𝜕𝜕
velocity, and
𝑖𝑖 𝜕𝜕𝜕𝜕𝑖𝑖 where is the
p is the+mean(𝜌𝜌𝑢𝑢 𝑖𝑖 density,𝜕𝜕𝜕𝜕 pressure.
𝑗𝑗
̅𝑖𝑖 𝑢𝑢 𝑖𝑖̅ 𝑗𝑗+𝜕𝜕𝑢𝑢 ̅̅̅̅̅̅̅
the
𝜌𝜌𝑢𝑢 ′𝜕𝜕𝑢𝑢 ̅
Reynoldsare
𝑗𝑗 −the mean
stress, 𝑢𝑢
̅
𝑖𝑖𝑖𝑖 is
viscous the averaged
Description
stress Cartesian
of
tensor condition component
components, NSTM
as 90–100 (5)
rating*
follows: k s (µm) Rt50 (µm)
𝑗𝑗 ) = + 𝑖𝑖 Medium calcareous
̅̅̅̅̅̅ 𝜏𝜏stress,
̅̅̅ = 𝑗𝑗𝜇𝜇(𝑢𝑢̅𝑖𝑖 isvelocity, Hydraulically Heavy calcareous
smoothfouling fouling 0. (6) 70–80 10,000 0 10,0003,000 0 3,0
𝜏𝜏+ )the
𝜌𝜌𝑢𝑢 ′ ′ 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕mean 𝜕𝜕𝜕𝜕 𝜕𝜕𝜕𝜕
𝑖𝑖 𝑢𝑢𝑗𝑗 is and
surface
density, velocity, averaged the CartesianReynolds
𝑖𝑖p is the 𝑖𝑖 component
𝑖𝑖𝑖𝑖 pressure.
𝜕𝜕𝜕𝜕𝑗𝑗 ̅̅̅
the are
𝑖𝑖𝑖𝑖𝜕𝜕𝜕𝜕
averaged and mean 𝑖𝑖pCartesian
isviscous
the mean
𝑗𝑗 Heavy component
stress
Typical astensor
calcareous*
applied NSTM components,
fouling
AF (2002):
coating Navalas follows:
Ships’ Technical Manual
0 90–100 30 10,000 150 10,
𝑖𝑖
̅̅̅̅̅̅
where 
the meanpressure.
where pressure.
is the thedynamic 𝜏𝜏density,
̅̅̅𝑖𝑖𝑖𝑖 are arethe themean mean
𝜌𝜌𝑢𝑢 ′ 𝑢𝑢viscous′viscous 𝜕𝜕𝑢𝑢̅𝑖𝑖 stress stress
𝜕𝜕𝑢𝑢 ̅𝑗𝑗tensor tensor components,
components, 𝑢𝑢̅
as follows:
𝑗𝑗𝜇𝜇(is𝜕𝜕𝑢𝑢̅the Reynolds stress, is(2002):
the averaged Cartesian
* Deteriorated
NSTM coatingNaval
or light Ships'
slimeTechnical Manual component 10–20 (6) 100 300
𝜇𝜇asis follows: 𝜏𝜏
̅̅̅
viscosity. 𝑖𝑖=
𝑖𝑖𝑖𝑖̅̅̅
𝜏𝜏𝑖𝑖𝑖𝑖 = 𝜇𝜇(𝜕𝜕𝜕𝜕𝑗𝑗 + 𝜕𝜕𝜕𝜕)𝑖𝑖
𝑖𝑖 + 𝜕𝜕𝑢𝑢 ̅ 𝑗𝑗 ) Heavy 𝑖𝑖 slime
The proposed roughness (6) function 30
model is as follows:
300 600
velocity, and 𝜕𝜕𝑢𝑢
p ̅𝑖𝑖is 𝜕𝜕𝑢𝑢 the ̅𝑗𝑗 mean pressure. 𝜕𝜕𝜕𝜕𝑗𝑗 ̅̅̅ 𝜏𝜏 𝜕𝜕𝜕𝜕
𝑖𝑖𝑖𝑖 𝑖𝑖 are the mean viscous
The
Small calcareous
proposed stress
(6) tensor
fouling
roughness
or weedcomponents, as 40–60
function model is as70–80 follows:
follows:
1,000 1,000
𝜏𝜏𝑖𝑖𝑖𝑖 = 𝜇𝜇(
̅̅̅ + )
where 𝜇𝜇where
The Reynolds-averaged is the 𝜇𝜇𝜕𝜕𝜕𝜕is𝑗𝑗dynamic
the 𝜕𝜕𝜕𝜕dynamic
𝑖𝑖 viscosity.
Navier-Stokes
viscosity. (RANS) method
Medium calcareous
is employed
Heavy calcareous
fouling
fouling to solve the governing 90–100
3,000
10,000
3,000
10,000
+
ynamic viscosity. 𝜕𝜕𝑢𝑢̅𝑖𝑖 𝜕𝜕𝑢𝑢 ̅𝑗𝑗 0 Technical Manual 𝑘𝑘𝑠𝑠 ≤ 2.5
equations The using Reynolds-averaged the commercial = 𝜇𝜇( CFD
𝜏𝜏Navier-Stokes
̅̅̅ + software )(RANS) STAR-CCM+. * NSTM (2002): Naval Ships'
The flow variables are discretised
+ (6)
The Reynolds-averaged 𝑖𝑖𝑖𝑖 Navier-Stokes (RANS) method method is (6)employed
is employedto solve 1 the
to governing
solve 𝜋𝜋 𝑙𝑙𝑙𝑙𝑙𝑙(𝑘𝑘𝑠𝑠 ⁄2.5)
the governing
in space
veraged using using
Navier-Stokes
equations second-order (RANS)
the commercial methodschemes, 𝜕𝜕𝜕𝜕 𝑗𝑗is employed
CFD 𝜕𝜕𝜕𝜕and
software 𝑖𝑖 theSTAR-CCM+.
toconvection
solve the term
governing
The
The proposed Δ𝑈𝑈of
flow + 𝑙𝑙𝑙𝑙
=the𝜅𝜅 turbulence
variables
roughness (0.2667
are discretised
function model
𝑠𝑠𝑠𝑠𝑠𝑠 [
𝑘𝑘𝑠𝑠+ ) isis used2 𝑙𝑙𝑙𝑙𝑙𝑙 (10)
as follows:
]
as a first-2.5 < 𝑘𝑘𝑠𝑠+ < 25 (7)
equations
where 𝜇𝜇in is using
theCFD the
dynamic commercial viscosity. CFD software STAR-CCM+. The flow variables
10 is used as+a first- are discretised
the
order commercial
upwind space using
scheme. software
second-order The STAR-CCM+. shearschemes, stress The the
and flow
transport variables
convection (SST) are discretised
term ofk-the  turbulence
turbulence model
𝑙𝑙𝑙𝑙 (0.2667
{𝜅𝜅 turbulence 𝑘𝑘𝑠𝑠 ) is applied to 𝑘𝑘the
+
𝑠𝑠 ≤ 2.5 𝑘𝑘 + ≥ 25
𝑠𝑠
in spaceorder
cond-order using
where schemes, μ second-order
upwind is the and dynamic
scheme.the convection schemes,
Theviscosity. shearterm stress ofand the
transportthe convection
turbulence (SST)is k- used term
 turbulence
as a first- of themodel1 is applied 𝜋𝜋is
to used
the
𝑙𝑙𝑙𝑙𝑙𝑙(𝑘𝑘𝑠𝑠 ⁄2.5) as a first-
+
complete
The RANS
Reynolds-averaged equations, as
Navier-Stokes this combines (RANS) the advantages
method A brief of
explanation
is the
employed
Δ𝑈𝑈 +
= k- 
of and
𝑙𝑙𝑙𝑙 the
(0.2667 k-ε + 𝑠𝑠𝑠𝑠𝑠𝑠 [ 2 𝑙𝑙𝑙𝑙𝑙𝑙 (10) ]
𝑠𝑠 ) turbulence
𝑘𝑘solve
development
to the of this models.
2.5 < 𝑘𝑘𝑠𝑠+ <
roughness
governing 25
function
(7)
model is presented in Ap
heme. The
order upwind
completeshear stress RANStransport
scheme. equations, The (SST) shear
as this  turbulence
k-stress
combines transport model(SST)
the advantages is applied of the k- toturbulence
k- the
and k-ε 𝜅𝜅 turbulence model models.is applied to the
In
equations this Inmodel, The
using the
Reynolds-averaged
the k-the
k-commercial formulation
 formulation Navier-Stokes
CFD isisthe applied
softwarek- and (RANS) to the
STAR-CCM+. A.
inner
method part ofA the
The
1 boundary
brief explanation + layer,
ofk-ε and
the development the k-ε of25
𝑘𝑘𝑠𝑠+ ≥ this roughness
equations,
complete as
this
RANSthismodel, combines the
equations, advantages
as this of
combines applied to the
the k-ε turbulence
inner
advantages part ofmodels. the
of boundary
the k-𝜅𝜅flow
{ 𝑙𝑙𝑙𝑙 (0.2667
layer,
and variables
𝑘𝑘and
k-ε 𝑠𝑠 ) the
turbulence are discretised
models.
k-
n space 
eformulation is
formulation employed
formulation to
using second-order the
is toapplied to
free
the solve
free stream,
to the the
stream, inner
schemes, governing
which
which part of
givesgives
theequations better
boundary
better
and the convection using
predictions predictions
layer, A the
andof
brief
Fig. 1 term the function
of
k-ε
flow
explanation
shows of the
separation
of
a schematic
the model
flow
the is
and
development presented
separation adverseof this in
and Appendix
roughness adversefunction A.model is presented
functioninmodel
Appendix
In
he thisstream,
free model,
commercial
pressure which thegives
gradients.
k- To
CFD
formulation
software
better achieve predictions STAR-CCM+.
reliable
isofapplied the
results, The
flow an
toflow
thevariables
separation
appropriate
inner
A.and choice
part
adverse of
of the1isturbulence
Fig.
grids
boundary
shows
representation
crucial. a schematic
is used
layer, of the
and asproposed
a first-
the
representation
k-ε roughness
of the proposed
in Eq. (
pressure
order upwind gradients. To achieve reliable results, an appropriate
compares it choice
k- turbulence with the of grids
measurements is crucial. of Schultz and Flack [29]. Good agreement is ob
formulation
ts. To achieve are discretisedtoscheme.
reliable theresults, free in space
The
stream, shear
an appropriate using which stress
second-order
choice gives transport
of gridsbetterschemes,
(SST)
predictions
is crucial. betweenand the of
roughness
two. the flow model
functionseparation is applied
model and
in Eq.
Fig. 1 shows a schematic representation of the proposed roughness function model in Eq. (7), and
to the
adverse
(7), and compares it with
theof
complete PhysicalRANS modellingequations, of the as roughness this combines geometry is the notadvantages
practical in CFD,the duek- to itsand complexity. The
pressure
Physical
ng
the convection
gradients.
modelling
of thewall-function
roughness geometry Toof termachieve
the is
of the
roughness
not practical
turbulence
reliable in results,
geometry
CFD,
isdue used an
is
to
ascomplexity.
appropriate
not
its
first- ofitThe
a practical
compares the
choice measurements
innear
with CFD, grids
due
measurements k-εto ofturbulence
isitsSchultz
crucial.
ofcomplexity.
Schultz and andmodels.
Flack
Flack [29].
The [29].GoodGoodagreement
agreement is observed
approach is therefore applied to solve the flow equations the wall, rather than
In this
wall-function
proach
order
model,
isusingtherefore
upwind the
approach k-
applied to is

scheme. formulation
solve
The
therefore the flow
shear is stress applied
applied transportto to
nearsolve
the (SST) inner k-ϖ
between
theratherflowthan
theis
part ofobserved
two. the between
boundary
equations near the wall, 7,9 the
layer, two. and
rather the k-ε
than
turbulence-model equations upequations to the wall. the wall,
Physical
formulation
-model
turbulence
equations
modellingto the
up
model
to free
the
ofwall. isstream,
the applied
roughness towhich the complete geometry
gives RANS
better is not equations, practical of
predictions in CFD,the flow dueseparation
to its complexity. Schultz and
and
Flack (2007)
adverse The
using turbulence-model
as this combines the equations
advantages ofupthetok-ϖ theand wall. k-ε turbulence
7,9 6,9 Proposed roughness function model
wall-function
pressure gradients. approach To achieve is therefore reliable applied results, to an solve the flow choice
appropriate equations near the
of6,9 grids is Proposedwall, rather than
Schultz
crucial. and Flack (2007)
models. In this model, the k-ϖ formulation is applied to the roughness function model
using turbulence-model
PROPOSED ROUGHNESS FUNCTION equations up to the wall. 5,9
Physical modelling inner part of the boundary layer, and
of the roughness geometry is not practical in CFD,5,9due to its complexity. Thethe k-ε formulation to
OUGHNESS FUNCTION
the free stream, which gives better predictions of the flow 4,9
PROPOSED
wall-function ROUGHNESS
approach is therefore FUNCTION appliedTo toachieve
solve reliable the flow equations 4,9 near the wall, rather than
separation and adverse pressure gradients.
In general, roughness equations functions areup ∆U+is+ no
using
PROPOSED turbulence-model
results, ROUGHNESS an appropriate choice of obtained grids to the experimentally,
wall.
is crucial. since there (m/s) universal
3,9 roughness
hness functions
function aremodel obtained for all kinds ofFUNCTION
experimentally, roughsince surfaces. there It is should
no universal be noted that∆U
roughness the (m/s)
impact 3,9
of roughness on
Physical modeling of the roughness geometry is not
or all kinds ΔU+ofdepends rough surfaces. on the type It shouldand coverage be notedofthat thethe impact of roughness on
roughness. 2,9
In general, practical roughness in CFD,functions due to its complexity. are obtained Theexperimentally,
wall-function since there 2,9
is no universal roughness
the type and coverage of the roughness.
Inapproach
this study, is therefore
the roughness applied parameters to solvemeasured the flow equations by Schultz near
and Flack [29] are 1,9used
1,9 to develop a
function model
PROPOSED for all kinds ofFUNCTION
ROUGHNESS rough surfaces. It should be noted that the impact of roughness on
In general,
roughness the roughness
+ newparameters
wall,
roughness rather than
function
measured functions usingmodel.
by Schultz turbulence-model
are Theobtained
and proposedFlack [29] model equations
experimentally,
are used is fitted toup to the
develop since a there is
roughness no universal
function values roughness
ΔU dependsto the on
wall. the type and coverage of the roughness. 0,90,9
function model for all kinds of rough surfaces. It should be noted that the impactitof
unction reported
model. by
The Schultz
proposed and Flack
model [29].
is fitted One to advantage
the roughness of our roughness
function values function model is that canroughness on
ultz and be
Flack applied[29]. to
One all types
advantage of fouled
of our surfaces
roughness and typical
function antifouling
model is that coatings.
it can The values of the sand
ΔU + -0,1
In
types depends
this
ofgrainstudy,
PROPOSED
fouled on
thethe
roughness
surfaces roughness type
ROUGHNESS
height
and typicaland and coverage
parametersother
antifouling FUNCTION relevant of
measured
coatings. the
dataroughness.
are by
The shown
values Schultz in the
of Table and1 Flack
sand for a range
-0,1
[29]
1
1 of aresurfaceused to develop 10 a
10
100
100
Innew
heightgeneral,
roughness
conditions.
and roughness
other function
relevant functions
datamodel. are shown areThe obtained Table 1experimentally,
in proposed formodel a rangeisoffitted since
surface to there
the roughness is no universal function roughness
+
ks (log scale)
values
+ (log scale)
In this study, thefor roughness parameters measured byofSchultz and Flack [29] are used to k
is develop ona the values reported by Schultz and
s
function
reported model
by In Schultz general, alland kinds
roughness
Flack of [29]. rough functionsOne surfaces. advantage are It should obtained our
Fig.be1.roughness
noted
Comparison thatof the the
function impact
proposed model
roughness of roughness that it can
Fig. 1. Comparison of the proposedfunction roughness model with
function model with the values
new
ΔU + roughness function model. The proposed modelFig.
be applied depends experimentally,
to on all the types since
type ofTablethere
and
fouled isRepresentative
1.coverage nosurfacesuniversal of coatingroughness
the
and roughness.
typical
and foulingfunction is fitted
1. Comparison
antifouling
conditions
to the
coatings.
[23]
roughness
of the proposed Thereported
function
roughness
values Flack
by Schultz [29]
offunction
and
values
theFlack model
sand[29] with the values reported by Schul
reported
grain Table byRepresentative
model
1.
roughness Schultz
for allheight kindsand of
coating Flack
rough
and andother [29].
surfaces.
fouling One
conditions
relevant It advantage
should [23]
data be noted are ofThe our
that
shown roughness
proposedinroughness Table function1 formodel
function
model
a range is of
has a similar
that
Flack it [29]
surface can
form to the built-in wall function of STAR-
In
be this study,
applied to the roughness
all types parameters
of fouled measured
surfaces by Schultz
and 4typical CCM+, andand
antifouling Flack as[29]
coatings.
is employed arefunction
theThe
wall used of
values toofdevelop
the sand
STAR-CCM+. a The proposed model is introduced
conditions. 4 The proposed
toshown
the fitted
roughness
CFD simulation setup
function
via
model hasB a= similar
theacoefficients 0,
form to the built-in wall function of
C = 0.2667, R+smooth = 2.5, and R+rough = 25
new
grainroughness
roughnessfunction model.
height and Therelevant
other proposeddatamodel is
areCCM+, in to the
Table roughness
1 for function
range of values
surface
and is employed as the wall function of STAR-CCM+. The proposed model is intr
reported
conditions.by Schultz and Flack [29]. One advantage ofto our the CFDroughness
simulationfunction
setup via model is that Bit=can
the coefficients 0, C = 0.2667, R+smooth = 2.5, and R+ro
be applied to all typesTable
of fouled surfaces andcoating
1. Representative typicaland
antifouling POLISH
coatings.
fouling conditions MARITIME RESEARCH,5
The values of the sand
[23] No 4/2023 63
grain roughness height and other relevant data are shown in Table 1 for a range of surface 5
Table 1. Representative coating
4 and fouling conditions [23]
conditions.
The proposed roughness function model has a similar form are 2D. The computational domain consists of two parts: the
to the built-in wall function of STAR-CCM+, and is employed inner region, which rotates with the propeller, and the outer
as the wall function of STAR-CCM+. The proposed model is part, which is fixed.
introduced to the CFD simulation setup via the coefficients
B = 0, C = 0.2667, R+smooth = 2.5, and R+rough = 25 (Eq. (7)),
to replace the STAR-CCM+ coefficients B = 0, C = 0.253,
R+smooth = 2.25, and R+rough = 90. Hence, the mathematical
formulation and flow calculations around the roughed
propeller are the same as for the smooth propeller except
for the roughness function model.
PROPELLER GEOMETRY
The VP1304 propeller has often been used for computational

case studies, and was selected here for analysis. It is a five-
blade right-handed propeller model with a diameter of 250 Fig. 3. Dimensions of the computational domain
mm (Fig. 2). Specifications of the propeller are shown in
Table 2 [30].
Tab. 2. Specifications of the VP1304 propeller Appropriate boundary conditions need to be applied to
ensure accurate simulations. In this study, we set the velocity
Parameter Symbol Value Units inflow for the inlet and the pressure boundary conditions for
Number of blades Z 5 - the outlet. The outer walls - are set to the slip wall condition,
Diameter D 0.250 m while a no-slip rough wall condition -is applied to the propeller,
Area ratio A E/A0 0.779 - shaft, and hub, to represent rough surfaces. The water density
Pitch ratio at 0.7 R P0.7/D 1.635 - and kinematic viscosity are 998.67 kg/m3 and 1.070×10-6 m/s2,
respectively. The boundary conditions are summarised in
Chord length at 0.7 R C0.7 0.104 m
Table 4 and Fig. 3.
Hub ratio dh/D 0.3 -
Tab. 3. Boundary conditions
Rotation rate n 15 rps
Advance coefficient J 0.6–1.2 - Region Boundary Type
Inflow speed VA adjust m/s Fixed parts Inflow Velocity inlet

Outflow Pressure outlet
Shaft No-slip condition
Outer walls Slip condition
Rotating parts Hub No-slip condition
Blades No-slip condition
Interface Nearest cell
interpolation
GRID GENERATION
In order to achieve reliable results, the generation of

appropriate grids is crucial. The rotating reference frame
(RRF) method is adopted for the propeller simulations in
Fig. 2. Propeller geometry this study. Since RRF does not require a complicated mesh
motion and a steady-state solver can be used, it is simpler
and computationally cheaper than unsteady approaches.
COMPUTATIONAL DOMAIN AND BOUNDARY In this approach, the domain remains stationary, with an
CONDITIONS assigned frame of reference rotating about an axis in the
global coordinate system. Fig. 4 depicts the structured and
The computational domain is depicted in Fig. 3, where unstructured grids, and Fig. 5 shows the propeller surface
the length and diameter of the domain for the open-water grids. The outer cylinder is meshed with coarse grids, and
simulations are 9D and 4D, respectively. The inlet is located at the inner cylinder with fine ones. The distance of the first
a distance 2D upstream, and the outlet is located at a distance cell from the wall, y+, is given in Fig. 6.
7D downstream, to avoid any reflections and to ensure
a uniform inflow. The top and bottom boundary distances

RESULTS
This section presents the CFD results for the propeller. The
computational results are first compared with open-water
experimental results for a smooth propeller, for validation
purposes. The simulation procedure is then applied to
a range of fouling conditions to investigate the effects of
fouling roughness on the open-water performance. Finally,
the required extra power is introduced and estimated in
terms of the specific sea margin, SSM, for the power loss of
the roughed propeller. Values of the SSM are obtained for
roughed and painted propellers.
MESH SENSITIVITY AND VERIFICATION

STUDY
Fig. 4. Domain grids A mesh sensitivity study is carried out to investigate

the thrust coefficient for coarse to fine grids. Fig. 7 shows
the uniform convergence of mesh sizes, and it can be seen
that there are no signs of divergence or oscillation. The grid
sensitivity is tested by estimating the numerical uncertainty.
The grid convergence index (GCI) with the Richardson
extrapolation [31] is employed to calculate the discretisation
error, and a mesh0,09
refinement factor r = √2 of is chosen.
0,08
0,09 0,07
0,08 0,09
0,06
0,07 0,08
0,09
Error
0,05
0,06 0,07
0,08
0,04
Error
0,05
0,06
0,07
0,04 0,03
ErrorError
0,05
0,06
0,03 0,02
0,02 0,04
0,05
0,01
0,01 0,03
0,04
0
0
0,03 0
0,02 0,5 1 1,5 2 2,5 3 3,5 4
0 0,5 1 1,5 2 2,5 3 3,5 4 4,5
0,01
0,02 Cell no. (mil)
Cell no. (mil)
Fig. 5. Propeller surface grids 0
0,01
Fig. 7.0 Grid convergence
0,5 1 results1,5
for KT 2 2,5 3 3,5 4
0
Fig. 7. Grid
0 convergence
Fig.
0,5 7. 1Grid 1,5 forCell
results KT2 no. (mil)
convergence results 3for
2,5 K3,5
T 4
he uncertainty is obtained Cellequations:

no. (mil)
The uncertainty is obtained using the followingusing the following
equations:
The uncertainty is obtained using the following equations:
𝜀𝜀21 = 𝜙𝜙2 − 𝜙𝜙1 Fig. 7. Grid convergence results for KT
𝜀𝜀32 = 𝜙𝜙3 − 𝜙𝜙2 𝜀𝜀21 = 𝜙𝜙2 − 𝜙𝜙1
𝜀𝜀32numbers
= 𝜙𝜙 Fig. 7. Grid convergence results for K
(8)
3 −following
𝜙𝜙 T
where the
Thesubscripts
uncertainty1, 2, oris3obtained
represent the using the assigned
2 toequations:
fine, medium, and coarse gr
where
respectively. 𝜙𝜙 is
thethesubscripts
function under 1, 2,consideration,
or 3 represent which the
in numbers
this case isassigned
KT or KQ. to fine, m
The uncertainty is obtained using 𝜀𝜀 = the−following
𝜙𝜙 𝜙𝜙 equations:
respectively. 𝜙𝜙 is the
where the𝑠𝑠 subscripts 𝜀𝜀
function under
21 2 consideration,
1 which in this case is
= 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 ( 1,) 2, or 3 represent the numbers assigned
32
𝜀𝜀 𝜀𝜀32 = 𝜙𝜙
𝜀𝜀21 =grids, 3 − 𝜙𝜙 2
𝜙𝜙2 −𝜀𝜀respectively.
𝜙𝜙1
to fine, medium, and21coarse is the function
32 the numbers assigned to fine, m
where theunder subscripts 1 1, 2,𝜀𝜀32orwhich
consideration, 𝑠𝑠𝜀𝜀3
= represent
32 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
= in (− 𝜙𝜙case
𝜙𝜙3this 2) is K or K . (1
𝑝𝑝𝑎𝑎 = |𝑙𝑙𝑙𝑙 |( )| + 𝑞𝑞(𝑝𝑝𝑎𝑎 )| 𝜀𝜀21 T Q
respectively. 𝜙𝜙 is 21the 1,
𝑙𝑙𝑙𝑙 (𝑟𝑟 )
where the subscripts 𝜀𝜀
function
2,21or 3 underrepresent consideration,
the numbers which in this
assigned tocase
fine,ism
𝑟𝑟21𝑎𝑎 −1𝑠𝑠 𝜀𝜀32
𝑝𝑝
respectively. 𝜙𝜙𝑞𝑞(𝑝𝑝 =𝑝𝑝𝑎𝑎𝑙𝑙𝑙𝑙function
is𝑎𝑎 )the =[ 𝑝𝑝𝑎𝑎 𝜀𝜀32)| + 𝑞𝑞(𝑝𝑝𝑎𝑎 )| which in this case is(1
|𝑙𝑙𝑙𝑙 |( consideration,
] under
𝑙𝑙𝑙𝑙 −(𝑟𝑟
𝑟𝑟21 =)𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 (𝜀𝜀21 )
𝑠𝑠𝑠𝑠21 (9)
𝜀𝜀𝜀𝜀𝑎𝑎21
𝑝𝑝
(𝑟𝑟21 −) 𝑠𝑠 ]
The extrapolated values are calculated as follows: 32
Fig. 6. Distance of the first cell from the wall, y+
𝑠𝑠 )==𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
𝑞𝑞(𝑝𝑝 1𝑎𝑎 𝑙𝑙𝑙𝑙 [ 𝜀𝜀𝑝𝑝𝜀𝜀32
𝑝𝑝(𝑟𝑟 𝑝𝑝
𝑎𝑎 =𝜙𝜙1 − 𝜙𝜙2 ) |𝑙𝑙𝑙𝑙 |(𝑟𝑟2121− 𝑎𝑎
)| 𝑠𝑠+ 𝑞𝑞(𝑝𝑝𝑎𝑎 )| (1
21
𝜙𝜙𝑒𝑒𝑒𝑒𝑒𝑒 = 21 𝑝𝑝 𝑙𝑙𝑙𝑙 (𝑟𝑟 1 21 ) 𝜀𝜀𝜀𝜀2132
The extrapolated values 𝑝𝑝𝑎𝑎 (𝑟𝑟 − 1)calculated
=21 are |𝑙𝑙𝑙𝑙 |( 𝑝𝑝𝑎𝑎 as )| + follows:
𝑞𝑞(𝑝𝑝𝑎𝑎 )|
The approximate and extrapolated relative 𝑙𝑙𝑙𝑙 (𝑟𝑟21 ) are𝑟𝑟21
errors 𝜀𝜀obtained
21
− 𝑠𝑠 as: (10)
𝑞𝑞(𝑝𝑝𝑎𝑎 ) = 𝑙𝑙𝑙𝑙𝑝𝑝 [ 𝑝𝑝𝑎𝑎 ]
21
(𝑟𝑟21 𝜙𝜙𝑟𝑟1 − 𝑎𝑎 − 𝜙𝜙2𝑠𝑠)
𝜙𝜙1 −𝜙𝜙𝜙𝜙𝑒𝑒𝑒𝑒𝑒𝑒
2 )== 𝑙𝑙𝑙𝑙 [𝑝𝑝 21
21
𝑒𝑒𝑎𝑎 = | 𝑞𝑞(𝑝𝑝 |𝑎𝑎 (𝑟𝑟21𝑟𝑟 𝑝𝑝−𝑎𝑎 as 1) ] (1
The extrapolated
POLISH valuesMARITIME 𝜙𝜙 are calculated
RESEARCH, 1 No𝑠𝑠follows:
− 4/2023
21 65
The
Theapproximate
extrapolated and 𝜙𝜙extrapolated
21 values
12
𝑒𝑒𝑒𝑒𝑒𝑒 − relative
𝜙𝜙1 calculated
are 𝑝𝑝 errors are obtained as:
as follows: (1
𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 =| | (𝑟𝑟 𝜙𝜙 − 𝜙𝜙 )
12 21 21 1 2
𝜙𝜙𝑒𝑒𝑒𝑒𝑒𝑒
𝜙𝜙𝑒𝑒𝑒𝑒𝑒𝑒 = 𝜙𝜙𝑝𝑝1 𝑝𝑝− 𝜙𝜙2
(𝑟𝑟21
𝜙𝜙1−−1)
The fine GCI is computed as: 𝑒𝑒𝑎𝑎21== (𝑟𝑟
21
𝜙𝜙𝑒𝑒𝑒𝑒𝑒𝑒 | 21 𝜙𝜙| 2 )
The approximate and extrapolated relative 𝜙𝜙
𝑝𝑝 1
21 (𝑟𝑟 − 1) errors are obtained as:
The uncertainty 𝜀𝜀32 = 𝜙𝜙 3 −following
𝜙𝜙2
respectively. 𝜙𝜙 isisthe
where the subscripts
obtained
function
1,
using
2, or 3 under
the
represent consideration,
the numbers
equations:
which in this case is KT or KQ.
assigned
The uncertainty is obtained using Fig.the 7. Grid
followingconvergence results fortoKTfine, medium, and coarse grids,
equations:
𝜀𝜀21 = 𝜙𝜙2 −𝜀𝜀32 𝜙𝜙1 (8)
respectively. 𝜙𝜙 is the function =under
𝑠𝑠𝜀𝜀32 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 consideration, which in this case is KT or KQ. (9)
= 𝜙𝜙3 (−𝜀𝜀 𝜙𝜙2)
The uncertainty is obtained 𝜀𝜀21 = 𝜙𝜙
using the2 −𝜀𝜀 𝜙𝜙1
following
21 equations: (8)
where the subscripts 1, 2, or𝑠𝑠 3=represent 32 the numbers assigned to fine, medium, and coarse grids, (9)
1𝜀𝜀32 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
= 𝜙𝜙3 (−𝜀𝜀32 𝜙𝜙2) (10)
respectively. 𝜙𝜙 is 𝑝𝑝𝑎𝑎function
the = 𝜀𝜀 = |𝑙𝑙𝑙𝑙
under 𝜙𝜙 |(− 𝜀𝜀21
𝜙𝜙 )| + 𝑞𝑞(𝑝𝑝𝑎𝑎 )| which in this case is KT or KQ.
consideration,
21
where the subscripts 1, 2,𝑙𝑙𝑙𝑙or(𝑟𝑟321 )represent 2 1
𝜀𝜀 the numbers assigned to fine, medium, and coarse grids, (8)
1𝜀𝜀32 = 𝜙𝜙3 −𝜀𝜀𝑝𝑝21 𝜙𝜙2
respectively. 𝜙𝜙 is the𝑝𝑝𝑎𝑎1,
=
function 32
𝜀𝜀32
𝑎𝑎 )| + 𝑞𝑞(𝑝𝑝 )|
|𝑙𝑙𝑙𝑙 |(consideration, VALIDATION . OF THE(10)
SIMULATIONS
where the subscripts 2, or𝑠𝑠 3=under
𝑙𝑙𝑙𝑙 𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠 (𝑟𝑟21 −)the
𝑞𝑞(𝑝𝑝(𝑟𝑟𝑎𝑎21
represent
)=
𝑠𝑠 numbers
) 𝑙𝑙𝑙𝑙 [ 𝜀𝜀𝑝𝑝21 𝑎𝑎 which in this
assigned to case
fine,ismedium,
]
KT or KQand (11)
(9)
coarse grids,
𝜀𝜀𝑝𝑝21𝑎𝑎
𝑟𝑟consideration,
(11)
respectively. 𝜙𝜙 is the function 𝜀𝜀32 𝑎𝑎 − 𝑠𝑠
under 𝑟𝑟 21 − 𝑠𝑠 which The open-water
in this case is KT or Ksimulation
Q. results(9)are compared with the
(11)
The extrapolated values 𝑞𝑞(𝑝𝑝𝑠𝑠1𝑎𝑎calculated
are )==𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
𝑙𝑙𝑙𝑙 [ ( 21 𝜀𝜀𝜀𝜀𝑝𝑝32as) follows: ] results of experiments performed (10)
by Barkmann et al. [32]
𝑝𝑝𝑎𝑎 = |𝑙𝑙𝑙𝑙 |(𝑟𝑟21 21
𝑎𝑎 )|
−+ 𝑠𝑠 𝑞𝑞(𝑝𝑝𝑎𝑎 )|
The extrapolated 𝑙𝑙𝑙𝑙 values
(𝑟𝑟
𝑠𝑠 =
21 )𝑠𝑠𝑠𝑠𝑠𝑠𝑠𝑠
are ( 𝜀𝜀𝜀𝜀21
calculated
32
) as follows: at a speed n = 15 rps for the smooth (9) propeller. The results
1calculated 𝑝𝑝 𝜀𝜀32 (10)
The extrapolated values 𝑝𝑝𝑎𝑎 = are (𝑟𝑟
|𝑙𝑙𝑙𝑙21|( 𝜙𝜙𝑟𝑟1𝜀𝜀𝑝𝑝21 −as
𝑎𝑎 )|
− 𝜙𝜙+ )𝑞𝑞(𝑝𝑝𝑎𝑎 )|
2follows:
𝑠𝑠 (12)
𝑙𝑙𝑙𝑙
21
(𝑟𝑟21 =) 𝑙𝑙𝑙𝑙 [ 𝑝𝑝21 𝜀𝜀21 for advance coefficients of J = 0.6–1.2
(11) are shown in Table 5.
𝑞𝑞(𝑝𝑝 1𝑎𝑎 ) = (𝑟𝑟 (𝑟𝑟
𝑝𝑝 21𝜀𝜀𝑝𝑝−
𝜙𝜙𝑟𝑟𝑟𝑟21𝑝𝑝 − 𝑎𝑎 1) ]
𝑠𝑠𝑠𝑠2 )𝑞𝑞(𝑝𝑝 )| (10)
1 𝑎𝑎−)| −𝜙𝜙+
32 Tab. 5. Experimental and simulation results for the open-water smooth propeller
𝑝𝑝𝑎𝑎 = 𝜙𝜙 21 = |𝑙𝑙𝑙𝑙 21|( 𝑎𝑎
(12)
(11)
The 𝑙𝑙𝑙𝑙 (𝑟𝑟𝑎𝑎calculated
) 𝑙𝑙𝑙𝑙relative 𝑝𝑝21 𝜀𝜀𝑝𝑝21as follows:
Theapproximate
extrapolated and extrapolated
values 𝑞𝑞(𝑝𝑝
are 𝑒𝑒𝑒𝑒𝑒𝑒 )=
21 [
(𝑟𝑟21𝑟𝑟 𝑝𝑝− 𝑎𝑎 1)
errors
] are obtained
(12) as:
Open-water test results [32] Open-water CFD results
𝑟𝑟− 21𝑎𝑎 − 𝑠𝑠
− 𝑠𝑠 errors are obtained as:
The approximate and extrapolated 𝜙𝜙 relative 21 𝜙𝜙 (11)
The extrapolated values are 𝑞𝑞(𝑝𝑝 )= = (𝑟𝑟
calculated|𝑙𝑙𝑙𝑙21[𝜙𝜙𝑟𝑟1𝑝𝑝𝑎𝑎−as
𝑝𝑝1 2 follows: ] J KT KQ ηO (13)KT KQ ηO
The approximate
𝑒𝑒𝑎𝑎𝑎𝑎21
21
and = extrapolated 𝜙𝜙21 1
−𝜙𝜙|𝑠𝑠2 ) relative errors are 0.6 0.6288 1.3964 0.4300
(12)
0.5961 1.3593 0.4188
𝑝𝑝
𝜙𝜙𝑝𝑝1 −−
(𝑟𝑟 𝜙𝜙1)
2 ) (13)
The extrapolated
obtained as: values are21 = (𝑟𝑟
calculated
𝑒𝑒21𝑎𝑎21 12 𝜙𝜙1−−as
21
|𝜙𝜙21𝑒𝑒𝑒𝑒𝑒𝑒 𝜙𝜙𝜙𝜙1|2follows: 0.8 0.5100 1.1780 0.5512 (12)
0.5056
(14) 1.1744 0.5481
The approximate and extrapolated 𝜙𝜙𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 == | 𝜙𝜙
relative
𝑝𝑝 1 | errors are obtained as:
𝑒𝑒𝑒𝑒𝑒𝑒 (𝑟𝑟
𝑝𝑝12𝜙𝜙
21
12 − 1) 1.0 0.3994 0.9749 0.6520 0.4008 0.9993 0.6383
21
(𝑟𝑟𝜙𝜙21𝑒𝑒𝑒𝑒𝑒𝑒
1−−𝜙𝜙𝜙𝜙 12
) (12)
(14)
The approximate and
The fine GCI is computed as: 𝜙𝜙 21 =
extrapolated
𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒
𝑒𝑒𝑒𝑒𝑒𝑒 = | 𝜙𝜙 −
relative
1 𝑝𝑝 12 2 𝜙𝜙 | errors are obtained as:1.2 0.2949 0.7760 0.7258 0.3007
(13) 0.7855 0.7311
𝑒𝑒𝑎𝑎21 = | (𝑟𝑟21 𝜙𝜙𝑒𝑒𝑒𝑒𝑒𝑒 − 1)| (13)
𝜙𝜙1
The fine
approximate and extrapolated 𝜙𝜙relative
1 − 𝜙𝜙221 errors are obtained as:
GCI is computed as:
𝑒𝑒𝑎𝑎2121
𝐺𝐺𝐺𝐺𝐺𝐺 = |𝜙𝜙121.25𝑒𝑒 − 𝜙𝜙𝑎𝑎1|
(13)
(15)
21 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 =𝑒𝑒𝑒𝑒𝑒𝑒 Fig. 8 shows both the numerical(14) and experimental results
𝑒𝑒𝑒𝑒𝑒𝑒𝑒𝑒 = | 𝜙𝜙 𝑟𝑟𝜙𝜙 −
𝑝𝑝1
12𝜙𝜙 −221 1 |
21
|𝜙𝜙are 𝜙𝜙
11.25𝑒𝑒
12 21 for the smooth propeller. In general,
(13)
(15)good agreement between
𝑒𝑒𝑒𝑒𝑒𝑒 −
21𝑒𝑒𝑎𝑎 21 =21 𝑒𝑒𝑒𝑒𝑒𝑒 𝜙𝜙𝑎𝑎|
a, q(pa), Φext , ea𝑒𝑒𝐺𝐺𝐺𝐺𝐺𝐺 ext| = (14)
21 1
where s, p 21 , e =
𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 𝜙𝜙 𝑝𝑝1intermediate | parameters. numerical and experimental results is achieved; a 5%
The fine GCI is computed as: 𝑒𝑒𝑒𝑒𝑒𝑒 𝑟𝑟21
𝜙𝜙 12 − 1 (14)
12
21𝜙𝜙𝑒𝑒𝑒𝑒𝑒𝑒 𝑒𝑒𝑒𝑒𝑒𝑒
− 𝜙𝜙1 discrepancy in K is the largest error observed for advance
where
The fine
GCIs, pGCI
a, q(p
is isa),
also Φext , ea𝑒𝑒as:
21
calculated
21
21 ,foreextKareand intermediate
K | . Valuesparameters.
for the numerical T
uncertainty 2.81%(14)
of0.6–1.2. and
𝑒𝑒𝑒𝑒𝑒𝑒 = | T 1.25𝑒𝑒
computed 21Q
21 12
𝜙𝜙𝑒𝑒𝑒𝑒𝑒𝑒 𝑎𝑎 coefficients in the range (15)
2.40% areisobtained 𝐺𝐺𝐺𝐺𝐺𝐺 𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 =
The GCI The fine
also for isthe
GCI
calculated discretisation
computed for K as:and 𝑝𝑝
𝑟𝑟21 −K 1errors
21
of KT and KQ, respectively. The required data and
Q. Values for the numerical uncertainty of 2.81% and
The fine GCI is computed as: 21
T 1.25𝑒𝑒 𝑎𝑎 (15)
details
where of
s, pthe GCI
q(p
2.40% area,obtained ), are
Φ presented
21
ea 𝐺𝐺𝐺𝐺𝐺𝐺
21
, e in
21
𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓
for, the discretisation =are Table 𝑝𝑝 4.
intermediate parameters.
a ext ext
𝑟𝑟21 − 21 1errors of KT and KQ, respectively. 10K
1,4 The required data and
1.25𝑒𝑒
Q
𝑎𝑎 1,2 (15)
details
where
The GCI s,ofpthe GCI
a, q(p
is also Φ
are
ext presented
21
a),calculated
21
, ea 𝐺𝐺𝐺𝐺𝐺𝐺 ,for 21
extin
21
K=are
e𝑓𝑓𝑓𝑓𝑓𝑓𝑓𝑓 Table
T and
4.
𝑝𝑝intermediate parameters. (15)
𝑟𝑟21 −K1Q. Values for the numerical uncertainty of 2.81% and
1
0,8
2.40% s,arepis
a,obtained Φextfor
21 the21discretisation 10
where
The GCI q(p a),calculated
also , ea ,for eext21KTareand Kerrors
intermediate of K
Q. Values
andtheKQnumerical
parameters.
Tfor , respectively.
K 0,6
The required
uncertainty data and
of 2.81%
T and
details of the
where
2.40% are obtained GCI
s, p , are
q(p
forpresented
), Φ 21
, ea 21
, in
e Table
21
are 4.
intermediate parameters.  0,4
a the ext discretisation errors 10
of K and K , respectively. The required data and
The GCI is also acalculated for KextT and KQ. Values Tfor the Qnumerical
O
0,2 uncertainty of 2.81% and
details of the GCI are presented in Table 4. 0
2.40% are obtainedThe GCI is also
for thecalculated
discretisationfor KT and KQ. of
errors Values for the
KT and KQ, respectively.
0,6 The
0,7 required
0,8 data
0,9 and 1 1,1 1,2
numerical uncertainty
details of the GCI are presented in Table 4. of 2.81% and 2.40% 10are obtained J
for the discretisation errors of KT and KQ, respectively. The KT (CFD) 10KQ (CFD) Eta (CFD)
10 in Table 4.
required data and details of the GCI are presented
KT (Exp.) 10KQ (Exp.) Eta (Exp.)
Fig. 8. Experimental [32] and CFD simulation results (current study) for a VP1304 propeller under
Tab. 4. Uncertainty calculations for KT and 10
KQ Fig. 8. Experimental
smooth open-water conditions [32] and CFD simulation results (current study) for
a VP1304 propeller under smooth open-water conditions
KT KQ The velocity contour at the surface of the propeller is shown in Fig. 9.
r21 √2 √2 The velocity contour at the surface of the propeller is shown

r32 √2 √2 in Fig. 9.
ϕ1 0.50560 0.11744
ϕ2 0.50304 0.11718
ϕ3 0.50095 0.11689
ε21 −0.00256 −0.00026
ε32 −0.00209 −0.00029
pa 0.58528 0.31508
ϕ ext21 0.51698 0.11969 Fig. 9. Velocity contour
e 21
a
0.50% 0.22%
EFFECT OF FOULING ON THE OPEN-WATER PERFORMANCE OF THE
e 21
ext
2.20% 1.88% PROPELLER
GCIfine21 2.81% 2.39%

In the following, the sand-grain roughness heights are used to represent the fouling conditions
(Table 1). The results for the propeller performance at a speed of n = 15 rps and values of J = 0.6–
Fig.
1.2 are presented for different fouling 9. Velocity
conditions in contour
Table 6 and Fig. 10, and are compared with
the results for the smooth propeller.
12

EFFECT OF FOULING ON THE OPEN-WATER
PERFORMANCE OF THE PROPELLER
0,8
0,7
0,6
In the following, the sand-grain roughness heights are used 0,5

O 0,4
to represent the fouling conditions (Table 1). The results for 0,3
the propeller performance at a speed of n = 15 rps and values 0,2
of J = 0.6–1.2 are presented for different fouling conditions 0,1

0,0
in Table 6 and Fig. 10, and are compared with the results for 0,6 0,7 0,8 0,9 1 1,1 1,2
the smooth propeller. J

Smooth propeller ks = 30 ks = 100 ks = 300
It can be seen that KT and ηo gradually decrease with ks = 1000 ks = 3000 ks = 10000
increasing fouling roughness, while KQ increases. For a value
of J = 1.2 for medium calcareous fouling, is reduced by about (c)
(c)
26% and KQ increases by about 7.0% with respect to the smoothFig. 10.Fig. Results for (a) the propeller thrust coefficient, (b) the torque coefficient, and (c) the efficiency
10. Results for (a) theforpropeller thrust coefficient, (b) the torque coefficient,
a range of surface conditions
propeller. Consequently, a 30.9% decrease in η is observed. The and (c) the efficiency for a range of surface conditions
It can be seen that KT and o gradually decrease with increasing ofouling roughness, while KQ
effects
increases. of medium
For a value of J = 1.2 and heavycalcareous
for medium calcareous fouling
fouling, on the
𝐾𝐾𝑇𝑇 is reduced by open-
about To
26%enable
and us to consider the roughness effect alone, Fig. 11(a) depicts KT as a function of ks. The
slope of KT versus ks indicates that there is a large decline up to ks = 500 μm over the whole range
water performance of the considered propeller are almost the
KQ increases by about 7.0% with respect to the smooth propeller. Consequently, a 30.9%
of J. It can To
decrease enable that
be observed us toKT consider the roughness
rapidly decreases as ks increases effect
up toalone,
ks = 1000 Fig.μm;11(a)
for ks in the
in o is observed. The effects of medium and heavy calcareous fouling on the open-water
same. The reason for this may be related to the relative
performance of the considered propeller are almost the same. The reason for this may be related
height depicts
range 1000–3000 Kμm,
T
asa a function
marginal of
decrease kof
s
. The
K T is slope
observed, of K
while
T
versus
K T tendsk s
indicates
to a constant for ks
of theheight
roughnessand the μm.is a large decline up to k = 500 μm over the whole
andsub-layer
the sub-layerthickness. that3000there
greater than
to the relative of the roughness thickness. s
0,7
When ηrange of asJ. aItfunction
O is plotted can be of observed
ks, it shows the sameKtendency
that T
rapidly as Kdecreases
T, as discussed k s (Fig.
asabove
11(c)). increases up to k = 1000 μm; for k in the range 1000–3000 μm,
0,6 s s
a marginal
Fig. 11(b) shows KQ as adecreasefunction of kof s. It K
can be is seen
observed, whilewith
that KQ increases KTkstends to of ks
up to a value
0,5 T
= 1000 μm. For ks in the range 1000–3000 μm, a marginal increase in KQ is observed, and KQ is
0,4 a constant for
constant for ks larger than 3000
k greater
s μm.
than 3000 μm.
KT
0,3 When ηO is plotted as a function of ks, it shows the same
The slopes for KT, KQ, and O versus ks are larger at a value of J = 1.2 than J = 0.6. Fig. 11(c) shows
0,2 tendency as K , as discussed above (Fig. 11(c)).
that for a higher value of J,T there is a more significant reduction in O. Figs. 10(c) and 11(c) also
0,1
show that a Fig.
slight 11(b)
increaseshowsin roughnessKQ as a function
leads of k s.inItthecan
to a large decrease valuebe of  seen that
O for the propeller.
0,0 This is K increases
conclusive
Q
evidence with thatk up
the
s
to
initiala value
roughness of k
up s
=
to 1000
small μm. For
calcareous k in
fouling
s
the
has crucial
range 1000–3000 μm, a marginalby Songincrease
et al. [25], in K is observed,
0,6 0,7 0,8 0,9 1 1,1 1,2
effect. This finding supports those of studies Farkas et al. [28], and Owen et
It can be seen that KT and o gradually decrease Jwith increasing fouling roughness, while KQ Q
al. [23], among others. Therefore, the importance of the initial roughness on the propeller
increases. For a value Smooth
of J propeller
= 1.2 for medium ks = 100 𝐾𝐾 is reduced
ks = 30 calcareous fouling,
𝑇𝑇
ks = by
300about 26% and and isKemphasised,
performance Q
is constant and
for k s larger
propeller painting
than
is
3000 μm.
recommended as a solution. In other words,
KQ increases by about ks =1000
7.0% with respect ks = 3000 ks = 10000
to the smooth propeller. Consequently, a 30.9% decrease The slopes for K K and η versus k are larger at a value of
the rate of required power will increase T, Q, as the roughness
O increases.
s The propeller roughness arises
in o is observed. The effects of medium and heavy calcareous fouling on the open-water from theJ accumulation
performance of the considered propeller are almost (a)
(a) the same. The reason for this may be related
= 1.2 than J
of= 0.6.
fouling Fig. 11(c)
as a function shows that
of the time for a
operatinghigher value ofthatJ, a large
at sea, meaning
to the relative height of the roughness and the sub-layer thickness.

drop
there is a more significant reduction in ηO. Figs. 10(c) and 11(c)
in propeller performance is expected in the early stages of operation.
1,4
0,71,2
also show that a slight increase in roughness leads to a large
0,61,0
decrease in the value of ηO for the propeller. This is conclusive
10KQ0,50,8 evidence that the initial roughness up to small calcareous
0,4
0,6 fouling has crucial effect. This 14 finding supports those of studies
KT 0,4
0,3
0,2
by Song et al. [25], Farkas et al. [28], and Owen et al. [23], among
0,20,0 others. Therefore, the importance of the initial roughness
0,1
0,6 0,7 0,8 0,9 1 1,1 1,2
on the propeller performance is emphasised, and propeller
J
0,0 painting is recommended as a solution. In other words, the rate
0,6 Smooth propeller ks 0,8
= 30 0,9 ks = 100 1 ks
1,1= 300
of required power will increase as the roughness increases. The
0,7 1,2
ks = 1000 ks = 3000 J ks = 10000
Smooth propeller ks = 30 ks = 100 ks = 300 propeller roughness arises from the accumulation of fouling
ks =1000 ks = 3000 (b) ks = 10000
as a function of the time operating at sea, meaning that a large
(b)
(a)
drop in propeller performance is expected in the early stages
of operation.
1,4
1,2
1,0
10KQ 0,8
0,6 13
0,4
0,2
0,0
0,6 0,7 0,8 0,9 1 1,1 1,2
J
Smooth propeller ks = 30 ks = 100 ks = 300
ks = 1000 ks = 3000 ks = 10000
(b)

13
Tab. 6. Computed open-water characteristics under different fouling conditions
Propeller surface/fouling
condition
J KT ΔKT (%) 10KQ ΔKQ (%) ηO ΔηO (%)
0.6 0.5961 0 1.3593 0 0.4188 0

0.8 0.5056 0 1.1744 0 0.5481 0
Smooth propeller
1.0 0.4008 0 0.9993 0 0.6383 0
1.2 0.3007 0 0.7855 0 0.7311 0
0.6 0.5889 −1.21 1.3630 0.27 0.4126 −1.48
0.8 0.4969 −1.71 1.1764 0.17 0.5378 −1.88
k s = 30 µm
1.0 0.3901 −2.66 0.9992 −0.01 0.6214 −2.65
1.2 0.2868 −4.62 0.7822 −0.42 0.7003 −4.21
0.6 0.5763 −3.33 1.3699 0.78 0.4017 −4.07
0.8 0.4832 −4.42 1.1825 0.69 0.5203 −5.07
k s = 100 µm
1.0 under different
Table 6. Computed open-water characteristics 0.3757 −6.27
fouling conditions 1.0063 0.70 0.5942 −6.92
ΔK (%)
1.2 0.2700
ΔK (%)
−10.20 0.7884 0.37 0.6541 −10.53
Propeller surface/fouling J K T 10K T Q Q O ΔO(%)
condition 0.6 0.5576 −6.47 1.3627 0.25 0.3907 −6.70
0.6 0.5961 0 1.3593 0 0.4188 0
0.8 0.5056 0 0.8 1.1744 0.4637
0 −8.28 0
0.5481 1.1789 0.38 0.5008 −8.63
Smooth propeller 1.0µm 0.4008
k s = 300 0 0.9993 0 0.6383 0
1.2 0.3007 0
1.0 0.7855
0.3523
0
−12.10 0
0.7311
0.9984 −0.10 0.5616 −12.01
0.6 0.5889 −1.211.2 1.3630 0.2412
0.27 −19.78−1.48
0.4126 0.7742 −1.44 0.5951 −18.60
0.8 0.4969 −1.71 1.1764 0.17 0.5378 −1.88
ks = 30 µm 1.0 0.3901 −2.660.6 0.9992 0.5317
−0.01 −10.81−2.65
0.6214 1.3725 0.97 0.3699 −11.66
−4.62 −0.42
1.2 0.2868
0.8 0.7822
0.4419 −12.60−4.21
0.7003
1.2042 2.54 0.4672 −14.77
0.6 µm 0.5763
k s = 1,000 −3.33 1.3699 0.78 0.4017 −4.07
0.8 0.4832 −4.421.0 1.1825 0.3553
0.69 −16.33−5.07
0.5203 1.0410 4.17 0.5127 −19.68
ks = 100 µm 1.0 0.3757 −6.27 1.0063 0.70 0.5942 −6.92
1.2 0.2700 −10.201.2 0.7884 0.2292
0.37 −23.77−10.53
0.6541 0.8380 6.67 0.5224 −28.54
0.6 0.5576 −6.470.6 1.3627 0.25
0.5279 −11.44−6.70
0.3907 1.3764 1.26 0.3663 −12.54
0.8 0.4637 −8.28 1.1789 0.38 0.5008 −8.63
ks = 300 µm 1.0 0.3523 0.8
−12.10 0.9984 0.4370
−0.10 −13.57−12.01
0.5616 1.2067 2.75 0.4611 −15.88
1.2 µm 0.2412
k s = 3,000 −19.78 0.7742 −1.44 0.5951 −18.60
0.6 0.5317 −10.811.0 1.3725 0.3295
0.97 −17.80−11.66
0.3699 1.0432 4.39 0.5026 −21.26
0.8 0.4419 −12.601.2 1.2042 0.2225
2.54 −26.02−14.77
0.4672 0.8408 7.04 0.5053 −30.89
ks = 1,000 µm 1.0 0.3553 −16.33 1.0410 4.17 0.5127 −19.68
1.2 0.2292 0.6
−23.77 0.8380 0.5279
6.67 −11.44−28.54
0.5224 1.3764 1.26 0.3663 −12.54
0.6 0.5279 −11.44 1.3764 1.26 0.3663 −12.54
ks = 3,000 µm 0.8 0.4370 0.8
−13.57 1.2067 0.4370
2.75 −13.57−15.88
0.4611 1.2067 2.75 0.4611 −15.88
1.0 µm0.3295
k s = 10,000 −17.801.0 1.0432 0.3295
4.39 −17.80−21.26
0.5026 1.0432 4.39 0.5026 −21.26
1.2 0.2225 −26.02 0.8408 7.04 0.5053 −30.89
0.6 0.5279 −11.441.2 1.3764 0.2225
1.26 −26.02−12.54
0.3663 0.8408 7.04 0.5053 −30.89
ks = 10,000 µm 0.8 0.4370 −13.57 1.2067 2.75 0.4611 −15.88
1.0 0.3295 −17.80 1.0432 4.39 0.5026 −21.26
1.2 0.2225 −26.02 0.8408 7.04 0.5053 −30.89
0,7 1,6
0,6 1,4
1,2
0,5
1,0
0,4
KT 10KQ 0,8
0,3
0,6
0,2
0,4
0,1 0,2
0,0 0,0
0 500 1000 1500 2000 2500 3000 0 500 1000 1500 2000 2500 3000
ks (µm) ks (µm)
J = 0.6 J = 0.8 J = 1.0 J = 1.2 J = 0.6 J = 0.8 J = 1.0 J = 1.2
(a) (b)
(a) (b)
0,8
0,7
15
0,6
0,5
O 0,4
0,3
0,2
68 POLISH MARITIME RESEARCH, No 4/2023 0,1

0,0
0 500 1000 1500 2000 2500 3000
ks (µm)
J = 0.6 J = 0.8 J = 1.0 J = 1.2

0,2
0,0
0 500 1000 1500 2000 2500 3000
ks (µm)
J = 0.6 J = 0.8 J = 1.0 J = 1.2
(b)
0,8
Roughness height for painted propellers
0,7
The roughness of a painted propeller over a period of
0,6
operation can be considered as the summation of the roughness
0,5
of the painted surface (k s1) and the accumulation of fouling
over a certain service time (k s2).
O 0,4
0,3
A value of k s1 = 0.17Ra can be used for painted propellers,
0,2
according to Schultz [33] (Table 8). It should be noted that
0,1
in this study, we use the roughness height of Schultz [33] but
0,0
not the corresponding roughness function model.
0 500 1000 1500 2000 2500 3000 Tab. 7. Annual roughness increments [34]
ks (µm)
Annual roughness increment
Coating type
J = 0.6 J = 0.8 J = 1.0 J = 1.2 Rt50 (µm/year)
(c) Traditional AF coating 40–60
(c) of (a) thrust coefficient, (b) torque coefficient,
Fig. 11. Results for propeller performance: graphs
Self-polishing coating, SPC 10–30
Fig. 11. Results for
andpropeller performance:
(c) efficiency graphs of height
versus roughness (a) thrust coefficient,
(b) torque coefficient, and (c) efficiency versus roughness height Foul-release paints, FR 5–15
LIDATION OF THE ROUGHED PROPELLER SIMULATIONS

The final roughness of the painted propeller is shown in
VALIDATION OF THE ROUGHED PROPELLER Table 8 for three types of paint after four years of service.
e literature does not contain a description of a rough propeller and a corresponding smooth
SIMULATIONS
peller as reference Tab. 8. Final roughness of a painted propeller after four years of service
that would enable a validation study, and it is therefore not possible to run
alidation study on the rough propeller. Although two papers have been published by Owen et
[23] and SongThe literature
et al. [25] thatdoes not contain
deal with rough anda smooth
description of a with
propellers, roughtwo different Roughness due to
Paint Final
fouling after four years
propeller
ghness function models,and a corresponding
insufficient smooth
data are publicly propeller
available as reference
to repeat these simulations. roughness
based on Table 1
roughness
that would enable a validation study, and it is therefore not
Paint type k s1 (µm) Rt50 (µm) k s2 (µm) k s (µm)
possible to run a validation study on the rough propeller.
Traditional AF
Although two papers have 16 been published by Owen et al. k s = 30 4×50=200 53 83
coating
[23] and Song et al. [25] that deal with rough and smooth
Self-polishing
propellers, with two different roughness function models, k s = 0.17Ra= 3.4 4×20=80 16 19.4
paints, SPC
insufficient data are publicly available to repeat these
Foul-release
simulations. k s = 0.17Ra= 2.4 4×10=40 8 10.4
paints, FR
It can be seen that for the same roughness conditions, the
Note:
results of both this study and prior works indicate a Self-polishing
significant paints, SPC ks = 0.17R a= SPC
3.4 and about 420=80 16
In general, R is about 20 µm for 14 µm for FR paints [33]
change in the hydrodynamic performance of the propeller. For paints, FR a ks = 0.17Ra= 2.4
Foul-release 410=40 8
example, Owen et al. [23] reported maximum changes Note:inInKgeneral,
T
, Ra is about 20 µm for SPC and about 14 µm for FR paints [33]
KQ, and ηO of −25.5%, +6.9%, and −30.3% for k s = 3,000 µm. Power estimation for fouled propellers
The results of the current study predict changes in KT, KQ, and The ship hull is assumed to be clean, and only the propeller
ηO of −26.0%, +7.0%, and −30.1% for k s = 3,000 µm. Power estimation forUnder
is fouled. fouled these conditions, the power of the fouled
propellers
On this basis, it can be seen that the changes in propeller propeller is derived based on the power for a clean propeller at
performance calculated in the current study are qualitatively
The ship hull the ship service
is assumed tospeed,
be clean, Vs. Forand a fouled
onlypropeller,
the propeller the required
is fouled. Under
similar to those of other research. torque, Q f, increases for a given delivered power, PD, which
power of the fouled propeller is derived based on the power for a clean propel
leads to a decrease in the propeller speed, n, and hence the
EFFECT OF PROPELLER PAINTING ONspeed, ENGINE Vs. Forship a fouled
speed, propeller,
Vs. To maintain the required
the ship speedtorque, Qf, increases
calculated for for a gi
BRAKE POWER P D , which leads to a decrease in the propeller
the clean propeller condition, the engine power must be speed, n, and hence the ship sp
the ship speedincreased
calculated (referred
for tothe here as PBf).propeller
clean The clean propeller
condition, scenario
the engine pow
Our results indicate the importance of surface(referred
conditionsto here is considered
as PBf). Thehere clean
as a benchmark
propeller to scenario
estimate the is extra power here as a b
considered
on fuel consumption, and hence greenhouse gasthe emissions,
extra power for the
forfouled propeller
the fouled that leadsthat
propeller to anleads
estimate to ofanSSM. The of SSM. T
estimate
which could be reduced by cleaning and appropriate propeller value of
clean propeller is calculated P B
for a clean propeller
as follows: is calculated as follows:
painting.
At the design stage, engine power is evaluated based on 𝑃𝑃𝐸𝐸 𝑅𝑅𝑇𝑇 𝑉𝑉𝑠𝑠
the assumption of smooth propeller conditions; however, 𝑃𝑃𝐵𝐵 = = (16)
a propeller will become rough, and the ship speed is 𝜂𝜂𝑂𝑂 𝜂𝜂𝐻𝐻 𝜂𝜂𝑅𝑅 𝜂𝜂𝑚𝑚 𝜂𝜂𝑂𝑂 𝜂𝜂𝐻𝐻 𝜂𝜂𝑅𝑅 𝜂𝜂𝑚𝑚
consequently reduced. To maintain the speed calculated
for the case of a clean propeller, one solution is to paintR
where the where
T is total RT resistance,
hull and m is and
is total hull resistance, the ηmechanical
m
is the mechanicalefficiency.
propeller to keep its roughness below a certain level. efficiency.
Assuming the same hull efficiency, mechanical efficiency and relative rotat
clean and fouledAssuming the same
propellers, hullthe
PBf for efficiency,
fouledmechanical
propeller efficiency
can be estimated as
and relative rotative efficiency for the clean and fouled
𝑃𝑃𝐸𝐸 𝑅𝑅𝑇𝑇 𝑉𝑉𝑠𝑠
𝑃𝑃𝐵𝐵𝐵𝐵 = =
POLISH MARITIME 𝜂𝜂𝑂𝑂𝑂𝑂RESEARCH,
𝜂𝜂𝐻𝐻 𝜂𝜂𝑅𝑅 𝜂𝜂𝑚𝑚 𝜂𝜂𝑂𝑂𝑂𝑂
No𝜂𝜂𝐻𝐻 𝜂𝜂𝑅𝑅 𝜂𝜂𝑚𝑚
4/2023 69
where O is the open-water efficiency of the fouled propeller.

PD, which leads to a decrease in the propeller speed, n, and hence the ship speed, Vs. To maintain
𝑃𝑃𝐸𝐸
𝑃𝑃 𝑅𝑅𝑇𝑇𝑉𝑉 𝑉𝑉𝑠𝑠
the ship speed calculated 𝑃𝑃 for =the clean
𝑃𝑃𝐵𝐵𝐵𝐵 =
𝐸𝐸 propeller= 𝑅𝑅condition,
=
𝑇𝑇 𝑠𝑠 (16)
the engine power must be increased(16)
𝜂𝜂 𝜂𝜂 𝜂𝜂
𝜂𝜂𝑂𝑂𝑂𝑂propeller 𝜂𝜂
𝜂𝜂𝐻𝐻𝐻𝐻𝜂𝜂𝑅𝑅𝑅𝑅𝜂𝜂𝑚𝑚 𝜂𝜂 𝜂𝜂 𝜂𝜂
𝜂𝜂𝑂𝑂𝑂𝑂𝜂𝜂𝐻𝐻𝐻𝐻𝜂𝜂is 𝜂𝜂
𝑅𝑅 𝜂𝜂considered
(referred to here as PBf). The clean 𝑚𝑚 scenario 𝑅𝑅 𝑚𝑚 here as a benchmark to estimate
𝑚𝑚
the extra power for the fouled propeller that leads to an estimate of SSM. The value of PB for a
clean
here RTpropeller
is total ishull
calculated and m is the mechanical efficiency.
here RT is total hull resistance, and m is the mechanical efficiency.
as follows:
resistance,
𝑃𝑃𝐸𝐸 𝑅𝑅𝑇𝑇 𝑉𝑉𝑠𝑠 (16)
ssuming the same hull 𝑃𝑃𝐵𝐵for
= the fouled
hullPBfefficiency,
thepropellers, efficiency, =propeller can
mechanical be estimated
efficiency andasrelative
calcareous
relative fouling,
rotative heavy for
efficiency slime fouling, antifouling paint,
for the
the
ssuming same 𝜂𝜂𝑂𝑂 𝜂𝜂𝐻𝐻 𝜂𝜂𝑅𝑅 𝜂𝜂𝑚𝑚 𝜂𝜂𝑂𝑂 𝜂𝜂𝐻𝐻 𝜂𝜂𝑅𝑅 𝜂𝜂𝑚𝑚 efficiency
mechanical and rotative efficiency
ean and follows:
fouled propellers,
ean and fouled propellers, PBf P for the fouled propeller can be
Bf for the fouled propeller can be estimated as follows: self-polishing
estimated as paint,
follows: and foul-release paint, respectively. It
where RT is total hull resistance, and m is the mechanical efficiency. can be seen that the painting of the propeller is extremely
𝑃𝑃𝐸𝐸
𝑃𝑃 𝑅𝑅𝑇𝑇𝑉𝑉
𝑅𝑅 𝑉𝑉𝑠𝑠 effective; for instance, the difference between SSM values of
(17)
Assuming the same hull𝑃𝑃 𝑃𝑃efficiency,
=
𝐵𝐵𝐵𝐵 =
𝐸𝐸
mechanical = efficiency
=
𝑇𝑇 𝑠𝑠
and relative 57.8%
rotative and
efficiency 2.7%foris
the(17)
huge in terms of fuel consumption and
𝐵𝐵𝐵𝐵 𝜂𝜂𝑂𝑂𝑂𝑂𝜂𝜂𝜂𝜂𝐻𝐻𝜂𝜂𝜂𝜂𝑅𝑅𝜂𝜂𝜂𝜂𝑚𝑚 𝜂𝜂𝜂𝜂𝑂𝑂𝑂𝑂𝜂𝜂𝜂𝜂𝐻𝐻𝜂𝜂𝜂𝜂𝑅𝑅𝜂𝜂𝜂𝜂𝑚𝑚 (17)
clean and fouled propellers, PBf𝜂𝜂𝑂𝑂𝑂𝑂 for the 𝐻𝐻 𝑅𝑅fouled𝑚𝑚 𝑂𝑂𝑂𝑂 𝐻𝐻 𝑅𝑅 can
propeller 𝑚𝑚 be estimated as gas emissions. Painting of propellers is therefore strongly
follows:
advised, using foul-release paint.
here 
here OO is
is the
the open-water
where ηO is the
open-water efficiency
open-water
efficiency of the
𝑃𝑃𝐸𝐸 efficiency
of the fouled
𝑅𝑅𝑇𝑇 fouled
𝑉𝑉
of
𝑠𝑠 the fouledpropeller.
propeller.
propeller. (17) of SSM values for fouled unpainted and fouled painted propellers
Tab. 9. Comparison
𝑃𝑃𝐵𝐵𝐵𝐵 = =
𝜂𝜂𝑂𝑂𝑂𝑂 𝜂𝜂𝐻𝐻 𝜂𝜂𝑅𝑅 𝜂𝜂𝑚𝑚 R 𝜂𝜂and
For the fouled propeller, 𝑂𝑂𝑂𝑂 𝜂𝜂𝐻𝐻η𝜂𝜂𝑅𝑅,𝜂𝜂𝑚𝑚are the same as for
T m
Propeller surface
or the
or the fouled
fouled thepropeller,
clean propeller.
propeller, RTT and
R Based
and on m ,, theare
are assumption
the
the same
same asηHfor
thatas for ηR clean
and the
the clean propeller.
Case propeller. Based on
Based
condition
on ks the
the(µm) ηOf OR ηO SSM (%)
where O isarethenot significantly
open-water changed
efficiency m
of due to the propeller
the fouled propeller.roughness,
 
PBf  and R as
ssumption
ssumption thatthat H and R are
are not significantly changed
not significantly changed due to the propeller due to the propeller
1
roughness,
roughness,
Small calcareous
P
foulingPBf is
Bf is
isHestimated follows: 1000 0.4634 57.8
stimated
stimated as follows:
For theasfouled propeller, RT and m, are the same as for the clean propeller. Based on the
follows: or weed
assumption that H and R are not significantly changed due to the propeller 2roughness, Heavy slime
PBf is 300 0.5544 31.9
𝜂𝜂
𝜂𝜂 𝑂𝑂
estimated as follows: 𝑃𝑃𝐵𝐵𝐵𝐵 = = 𝑃𝑃𝑃𝑃𝐵𝐵 𝑂𝑂 3 AF paintings (18)(18) 83 0.6310 15.9
𝑃𝑃 𝐵𝐵𝐵𝐵 𝐵𝐵 𝜂𝜂𝑜𝑜𝑜𝑜
(18)
𝜂𝜂𝑜𝑜𝑜𝑜 4 Self-polishing coatings,
𝜂𝜂𝑂𝑂 19.4 0.6974 4.8
𝑃𝑃𝐵𝐵𝐵𝐵 = 𝑃𝑃𝐵𝐵 SPC(18)
𝜂𝜂𝑜𝑜𝑜𝑜
Specific sea margin 5 Foul-release paints, FR 10.38 0.7116 2.7
pecific sea
pecific sea marginmarginSea margin is defined as the extra power required due to the Clean propeller (J=1.2) 0 0.7311 0
ea margin
Specific is
sea sea state in as
defined
margin comparison
the extra with power still water, as welldue
required as atoroughed
the sea state in comparison with still
ea margin ishull defined
in comparison
as the extra withpower
power hull
smooth
required
conditions.
due to define
the sea state in comparison with still
Seaas
ater, margin
well is defined
as a roughed as the extra
hull in comparison requiredwith to theTo
due smooth seahull
state in comparison
conditions. with
To define
definestillthe propeller
ater, as well the as apropeller
roughedroughness hull in comparison
effect onwith withpower,
engine smooth we hull
defineconditions. To the propeller
CONCLUSION
water, as
oughness effectwell
effect as a roughed
on engine
engine hull in comparison
power,(SSM) we define define smooth
the hull
‘specific conditions. To
sea margin’ define
margin’ (SSM) the
(SSM) as propeller
as the
the extra
extra power
power
oughness the ‘specific
on sea margin’
power, we as the extra
the power
‘specific
roughness effect on engine power, we define the ‘specific sea margin’ (SSM) as the extra power required
sea
equired
quired to
required to maintain
to tomaintain
maintain
maintain thethe
the
the shipspeed
ship
ship
ship speedininin
speed
speed inthethe
the casecase
the
case of
of aaroughed
case of
of propeller
aa roughed
roughed
roughed
propeller The compared
propeller
propeller
compared main
comparedgoal of
to a smooth
tothis
to study is to consider the effects
aa smooth
smooth
ropeller. compared to a smooth propeller. of the roughness of a propeller on its performance using
ropeller.
propeller.
The SSM is estimated as follows: the CFD method. A five-blade propeller model is selected
heThe
he SSM
SSMSSM is estimated
is estimated
is estimated as as follows:
as follows:
follows: for the calculations, and a simulation of a clean propeller
is verified. A new roughness function model is suggested
𝑃𝑃𝑃𝑃𝐵𝐵𝐵𝐵
𝐵𝐵𝐵𝐵 −− 𝑃𝑃𝑃𝑃𝐵𝐵 𝑃𝑃𝐵𝐵𝐵𝐵𝑃𝑃 𝜂𝜂𝑂𝑂 𝜂𝜂𝑂𝑂× 100 (19) (19)
100 based on existing measured (19) roughness data. The simulations
𝑆𝑆𝑆𝑆𝑆𝑆 =𝑃𝑃 − 𝑃𝑃𝐵𝐵 × 100 = ( 𝑃𝑃 −
𝐵𝐵𝐵𝐵 1) × 100 = ( − 𝜂𝜂
1)
𝑆𝑆𝑆𝑆𝑆𝑆 =
𝑆𝑆𝑆𝑆𝑆𝑆 = 𝐵𝐵𝐵𝐵 𝑃𝑃𝐵𝐵 𝐵𝐵 × × 100
100 ==𝑃𝑃((𝐵𝐵 𝐵𝐵𝐵𝐵 − − 1)
1) ×× 100
100𝜂𝜂𝑜𝑜𝑜𝑜=
= (( 𝑂𝑂 − − 1)
1) ×× 100 (19)
𝑃𝑃
𝑃𝑃𝐵𝐵𝐵𝐵 𝑃𝑃
𝑃𝑃𝐵𝐵𝐵𝐵 𝜂𝜂
𝜂𝜂𝑜𝑜𝑜𝑜
𝑜𝑜𝑜𝑜 are extended to represent the same propeller under several
roughness conditions, and the following conclusions could
The justification of this formula18SSM is examined and be drawn:
compared with ΔPs reported by Song et al. 18
18 [25]. The concepts –– A comparison of roughed and smooth propellers shows that
of SSM and ΔPs are similar, which is why these two different KT and ηo gradually decrease with increasing roughness up
methods show good agreement. to k s = 3,000 µm, while KQ increases up to k s = 3,000 µm.
For instance, at J = 1.2 for medium calcareous fouling,
Specific sea margin for painted propellers KT reduces by about 26%, KQ increases by about 7.0%, and
The performance of a propeller becomes worse with the ηO decreases by 30.9%.
fouling severity, as shown in Table 6. This table also shows –– The effects of medium and heavy calcareous fouling on
that the performance of a propeller coated with antifouling the open-water performance of the propeller are found to
paint is considerably better than that of a fouled propeller. The be almost the same.
effects of fouled painted propellers are examined based on the –– Graphs of KT, KQ, and ηO versus k s have larger slopes
propeller power, and SSM values for a propeller coated using at J = 1.2 than J = 0.6. A higher value of J gives a more
three types of paint, after four years of service (becoming significant reduction in ηO.
fouled), are considered. For the sake of simplicity, only the –– A slight increase in roughness leads to a large decrease in
open-water efficiency is considered when predicting the the value of ηo for the propeller. A large drop in propeller
extra power in terms of the SSM, and the effects of the other performance can therefore be expected in the early stages
parameters such as ηH, ηR, etc. are disregarded. of its operation.
The final roughness height, k s, for each type of paint after –– The painting of a propeller is extremely effective. For
four years of service is determined as shown in Table 8. The instance, a propeller coated with foul-release paint had
efficiency of a roughed propeller, ηof, is interpolated from an SSM of 2.7%, while the unpainted propeller required
Table 6 based on the final roughness height, and the SSM is 57.8% extra power with respect to the clean propeller for
calculated using Eq. (19). These data are presented in Table 9. a period of four years in seawater. Propeller painting using
It is worth mentioning that the first two cases represent fouled foul-release paint is therefore strongly advised.
unpainted propellers, and the remainder are fouled painted
propellers. At a given ship speed for a clean propeller, the SSM
is calculated as 57.8%, 31.9%, 15.9%, 4.8%, and 2.7% for small

Our recommendations for future research work are as –– A simulation of a full-scale fouled propeller could be
follows: a subject for future work, since the size of the roughness
–– The propeller model under roughed conditions could be cannot be scaled up from a model to a full-scale propeller.
tested in a towing tank. –– The model of roughness accumulation is based on an
–– The even distribution of roughness considered here could annual roughness increment, whereas measurements
be changed to a non-uniform real distribution, and both at shorter intervals (such as six months) on a full-scale
experiment and simulation could be conducted. propeller would be very helpful if practically possible.
NOMENCLATURE
A Roughness constant Uτ Friction velocity
B Smooth wall log-law intercept U +
Non-dimensional velocity
C, Cs Roughness constant ΔU+ Roughness function
D Diameter of propeller Vs Speed of the ship
GCI Grid convergence index VA Propeller advance speed
J Advance coefficient w Wake parameter
ks Equivalent sand-grain roughness height y Normal distance from the wall
k s+ Roughness Reynolds number based on ks y+ Non- dimensional distance from wall
KT Thrust coefficient ρ Fluid density
KQ Torque coefficient ϕκ KT and KQ on the kth grid
n Rotational speed of propeller µ Dynamic viscosity
PB, PBf Engine power (smooth, fouled) κ von Karman constant
PD Delivered power τij Mean viscous stress tensor components
PE Effective power ηD Quasi-propulsive efficiency coefficient
Q, Q f Propeller torque (smooth, fouled) ηH Hull efficiency
Ra, Rt50 Roughness height parameters ηm Mechanical efficiency
RT Total hull resistance ηO, ηOf Open-water efficiency (smooth, fouled)
SSM Specific sea margin ηR Relative rotative efficiency
t Thrust deduction factor ν Kinematic viscosity
T Propeller thrust ρui 'uj ' Reynolds stresses
ui Averaged Cartesian component velocity Uτ Friction velocity
5. P. Król, “Blade section profile array lifting surface design

REFERENCES method for marine screw propeller blade,” Pol. Marit. Res.
, 4 (1040), vol. 26, pp. 134-141, doi: 10.2478/pomr-2019-
1. IMO, Second IMO GHG Study 2009. London, UK, 2009. 0075, 2019.
2. A. Banawan, M. Mosleh, and I. Seddiek, “Prediction of the 6. A. Nadery and H. Ghassemi, “Numerical investigation of
fuel saving and emissions reduction by decreasing speed of the hydrodynamic performance of the propeller behind the
a catamaran,” J. Mar. Eng. Techno., (3), vol. 12, pp. 40-48, ship with and without WED,” Pol. Marit. Res. , 4 (108), vol.
2013. 27, pp. 50-59, doi: 10.2478/pomr-2020-0065, 2020.
3. M. H. Ghaemi and H. Zeraatgar, “Impact of propeller 7. M. Perić, “Prediction of cavitation on ships,” Brodogradnja,
emergence on hull, propeller, engine, and fuel consumption vol. 73, no. 3, pp. 39-58, 2022.
performance in regular waves,” Pol. Marit. Res. , (116), vol.
29, pp. 56-76, doi: 10.2478/pomr-2022-0044, 2022. 8. W. Tarełko, “The effect of hull biofouling on parameters
characterizing ship propulsion system efficiency,” Pol.
4. M. Burak Samsul, “Blade cup method for cavitation Marit. Res. , 4(84), vol. 21, pp. 27-34X, 2014, doi: 10.2478/
reduction in marine propellers,” Pol. Marit. Res. , 2 (110), pomr-2014-0038.
vol. 28, pp. 54-62, 10.2478/pomr-2021-0021, 2021.

9. L. Guangnian, Q. Chen, and Y. Liu, “Experimental a plate with inhomogeneous roughness,” Ocean Eng., vol.
study on dynamic structure of propeller tip vortex,” Pol. 237, p. 109628, 2021.
Marit. Res. , 2 (106), vol. 27, pp. 11-18, 2020, doi: 10.2478/
pomr-2020-0022. 22. P. Kellett, K. Mizzi, K. Demirel, and O. Turan,
“Investigation the roughness effect of biofouling on
10. Y. Zhang, X. Wu, M. Lai, G. Zhou, and J. Zhang, “Feasibility propeller performance,” International Conference on
study of RANS in predicting propeller cavitation in behind- Shipping in Changing Climates, SEMANTIC SCHOLAR,
hull condition,” Pol. Marit. Res. , 4 (108), vol. 27, pp. 26-35, ID 55216316, 2015.
2020.
23. D. Owen, Y.K. Demirel, E. Oguz, T. Tezdoga, and A.
11. ITTC Virtual, The Resistance and Propulsion Committee, Incecik, “Investigating the effect of biofouling on propeller
Final Report and Recommendations to the 29th ITTC, 13-18 characteristics using CFD,” Ocean Eng., vol. 159, pp. 505-
June ITTC 2021 Virtual. 516, 2018.
12. R. Townsin, and S. Dey, “The correlation of roughness 24. S. Song, Y. Demirel, and M. Atlar, “An investigation into
drag with surface characteristics,” Fluid Mechanics the effect of biofouling on full-scale propeller performance
and Its Applications, vol 6. Springer, Dordrecht, doi. using CFD,” in: Proceedings of the ASME 2019 38th
org/10.1007/978-94-011-3526-9_10, 1991. International Conference on Ocean, Offshore & Arctic
Engineering OMAE2019, June 9-14, 2019, Glasgow,
13. Y. K. Demirel, “A CFD model for the frictional Scotland, UK, 2019.
resistance prediction of antifouling coatings,” Ocean
Eng., vol. 89, pp. 21-31, 2014. 25. S. Song, Y. Demirel, and M. Atlar, “Propeller performance
penalty of biofouling: CFD prediction,” J. Offshore Mech.
14. A. Vargas, and H. Shan, “Modeling of ship resistance Arct. , vol. 142, no. 6, p. 0601901, 2020.
as a function of biofouling type, coverage, and spatial
variation,” in 2nd Hull Performance & Insight 26. A. Farkas, N. Degiuli, and I. Martić, “The impact of
Conference, pp. 264-281, Ulrichshusen, Germany, biofouling on the propeller performance,” Ocean Eng.,
2017. vol. 219, p. 108376, 2021.
15. C. Grigson, “Drag losses of new ships caused by hull finis,” 27. A. Farkas, N. Degiuli, and I. Martić, “Assessment of the
J. Ship Res., vol. 36, no. 02, pp. 182-196, 1992. effect of biofilm on the ship hydrodynamic performance
by performance prediction method,” International Journal
16. S. Song et al., “Investigating the effect of heterogeneous of Naval Architecture and Ocean Eng., vol. 13, pp. 102-
hull roughness on ship resistance using CFD,” JMSE, vol. 114, 2021.
9, no. 2, p. 1, 2021.
28. A. Farkas, S. Song, N. Degiuli, I. Martić, and Y.K. Demirel,
17. T. Cebeci and P. Bradshaw, “Momentum transfer “Impact of biofilm on the ship propulsion characteristics
in boundary layers,” in Series in Thermal and and the speed reduction,” Ocean Eng., vol. 199, p. 107033,
Fluids Engineering, P. Bradshaw, Ed. Washington: 2020.
Hemisphere Pub. Corp., 1977.
29. M. P. Schultz and K. A. Flack, “The rough-wall turbulent
18. A. Farkas, N. Degiuli, and I. Martić, “Towards the boundary layer from the hydraulically smooth to the fully
prediction of the effect of biofilm on the ship resistance rough regime,” J. Fluid Mech., vol. 580, pp. 381-405, 2007.
using CFD,” Ocean Eng., vol. 167, pp. 169-186, 2018.
30. SVA Hydrodynamic Solutions, “Potsdam Propeller Test
19. A. Farkas, N. Degiuli, L. Martic, and R. Dejhalla, “Impact Case (PPTC) Open Water Tests with the Model Propeller
of hard fouling on the ship performance of different ship VP1304,” 2011.
forms,” JMSE, vol. 8, p. 748, 2020.
31. L. F. Richardson, and J. A. Gaunt, “The deferred approach to
20. A. Farkas, N. Degiuli, and I. Martić, “Impact of biofilm the limit,” Philosophical Transactions of the Royal Society
on the resistance characteristics and nominal wake,” of London: Series A, vol. 226, pp. 299-361, 1927.
Proceedings of the Institution of Mechanical Engineers,
Part M: Journal of Engineering for the Maritime 32. U. Barkmann, H. Heinke, and L. Lubke, “Potsdam
Environment, vol. 234, no. 1, pp. 59-75, 2020. propeller test case (PPTC) test case description,” in: Second
International Symposium on Marine Propulsors SMP’11,
21. A. Farkas, N. Degiuli, and I. Martić, “A novel method for Hamburg, Germany, Workshop: Propeller Performance,
the determination of frictional resistance coefficient for 2011.

L. P.
SVA F.Hydrodynamic
Richardson, and J. A.“Potsdam
and K.Solutions, Gaunt, “The deferred
Propeller Test Caseapproach
(PPTC)tolayer the from
Open limit,”
Water Philosophical
theTests with the
elM. Schultz
Propeller VP1304,” A. Flack,
2011. “The rough-wall turbulent boundary hydraulically
sactions
el Propeller of the Royal
VP1304,” Society
2011. of London: Series A, vol.
oth to the fully rough regime,” J. Fluid Mech., vol. 580, pp. 381-405, 2007.226, pp. 299-361, 1927.
L. F. Richardson, In the rough regime, for limit,”
a roughness different from the sand-grain, the rou
.U.P.Barkmann,
L.
SVA Schultz and
F.Hydrodynamic H.K.
Richardson, A.and
Heinke,
and
J. A.
Flack,
and
J. A.
Solutions, “The
L.
Gaunt,
Gaunt,
“Potsdam
“The turbulent
rough-wall
Lubke, “Potsdam
“The
deferred
propeller
deferred
Propeller Test
approach
boundary
test
Case case
approach
(PPTC)
to the
layer
(PPTC)
to fromtest
the
Open thecase
limit,”
Water
Philosophical
hydraulically
description,”
Philosophical
Tests with the rough surface, while C is indiv
sactions
to
Secondthe of the
fully Royal
rough
International SocietyJ.of
regime,”
Symposium London:
Fluid on Mech.,Series
Marine vol. A, different.
580,vol.
pp.
Propulsors 226, Hence,
pp.
381-405,
SMP’11, 299-361,
2007.
Eq. (A.5)
Hamburg,
is applicable to any
1927.Germany, Workshop:
sactions of the Royal
del Propeller VP1304,” 2011. Society of London: Series A, vol. 226, pp. 299-361, 1927.
eller
U. Performance,
Barkmann, 2011. and L. Lubke, “PotsdamSchultz
H. Heinke, propeller and
test Flack
case Open [17] Water
(PPTC) introduced
test case awith
relation
description,” between an engineering surface
VAU.L. Hydrodynamic
Barkmann,
F. H.
Richardson, Solutions,
Heinke,and and
J. “Potsdam
L.
A. Lubke,
Gaunt, Propeller
“Potsdam
“The Test Caseapproach
propeller
deferred (PPTC)
test case (PPTC)
to the test TestsPhilosophical
case
limit,” the
description,”
econd International Symposium onofMarine Propulsors SMP’11,
sand-grain Hamburg,
roughness Fluidsbased Germany,
on the Workshop:
root meana relation
square between
of the roughness hei
M.
Propeller
Second
nsactions 33.
P. Schultz, M.
theP.Royal
VP1304,” Schultz,
“Frictional
International
of 2011.“Frictional
resistance
Symposium
Society resistance
on Marine
of London: of antifouling
antifouling coating
Propulsors
Series coating
systems,”
SMP’11,
A, vol. 226,
J. Schultz
Hamburg,
pp. 299-361, and
Eng.,
1927. Flack
vol.
Germany, [17]Workshop:
126, introduced
pp. 1039- an
eller Performance, 2011.
systems,”2011.
J. Fluids Eng., vol. 126, pp. 1039-1047, 2004.of the probability
flatness engineering surface density andfunction
an equivalent
(pdf).sand-grain
The mean roughness
velocity profile is
7,eller
2004.Performance,
based on the root mean pp. 1039- smooth for ks ≤ 2.5 (ΔU = 0
square of the roughness height,
.U.F.Barkmann,
Richardson, and J. A.
H. Heinke,resistance Gaunt,
and L. Lubke, “The deferred
“Potsdam and approach
it systems,”
can test
propeller be seen to
case the
that
(PPTC) limit,”
the flow
test Philosophical
is126,
case hydraulically
description,” + +
M. P. Schultz, “Frictional of antifouling coating J. Fluids Eng., vol.
ctions
Second of34.
J. Carlton,
M. P. Schultz, J.Royal
Carlton,
theMarine Marine
Propellers
“Frictional
Society ofPropellers
and
resistance
London: and Propulsion.
Propulsion.
onof Marine London:
antifouling
Series A, coating
vol. London:
Butterworth
226, systems,”
pp. skewness,
ks299-361,
+
≥ 25, Heinemann,
J. Fluids
1927. and
Eng., flatness
2010.
vol. 126, ofWorkshop:
the
pp. probability
1039- densityand function
ΔU+ will be a lin
, 2004. International Symposium Propulsors
velocity SMP’11, Hamburg,
the flow Germany,
regime becomes fully rough,
, 2004.
peller Butterworth
Performance, Heinemann, 2010.
2011.and L. Lubke, “Potsdam propeller (pdf). The
+ mean velocity profile is shown in Fig. A.1, and
Barkmann, H. Heinke, logarithmic scale
test case it of kbe
(PPTC)
can s .test
seen case
that description,”
the flow is hydraulically smooth for
J. Carlton, Marine Propellers and Propulsion. London: Butterworth Heinemann, 2010.
ond
J. P.International
M.Carlton, Marine
Schultz, Symposium
Propellers
“Frictional onPropulsion.
and
resistance Marine Propulsors
London:
of antifouling SMP’11,
Butterworth
coating systems,” kHeinemann,
Hamburg,
J.
s
+
≤ 2.5
Fluids Germany,
(ΔU + 2010. Workshop:
Eng.,= 0). For
vol. a
126, higher
pp. flow
1039- velocity k s
+
≥ 25, the
er 2004.
7, Performance, 2011. APPENDIX APPENDIX
A A flow regime becomes fully rough, and ΔU+ will be a linear
function of the logarithmic scale of k s+.
P. Carlton,
Schultz, Marine
“Frictional resistance
andofPropulsion.
antifoulingLondon:
coating systems,” J. Fluids Eng., vol. 126, pp. 1039-
J. Propellers APPENDIX A
Butterworth Heinemann, 2010.
004. PROPOSED ROUGHNESS FUNCTION APPENDIX MODEL A
OPOSED ROUGHNESS FUNCTION MODEL
Carlton, Marine Propellers and Propulsion. London: Butterworth Heinemann, 2010.
In this study, the roughness parameters measured by
OPOSED Schultz and Flack [29]
ROUGHNESS
is study, the roughness are
FUNCTION
parameters APPENDIX
utilisedmeasured
to develop
MODELaby A and Flack [29] are utilised to develop
newSchultz
roughness
OPOSED ROUGHNESS
function FUNCTION
model. The velocity profile onMODEL
a rough flat plate can
w roughness function model. The velocity profile on a rough flat plate can be expressed as
be expressed as follows: APPENDIX A
is study, the roughness parameters measured by Schultz and Flack [29] are utilised to develop
ows:
is study, the roughness parameters measured by Schultz and Flack [29] are utilised to develop
OPOSED
w roughness ROUGHNESS
function model. FUNCTION The 1velocity MODEL profile on a rough flat plate can be expressed as
w roughness function +model. 1 The
+ velocity + profile on a rough flat plate can be(A.1) expressed as
ws: 𝑈𝑈 = 𝑙𝑙𝑙𝑙 𝑦𝑦 − 𝑙𝑙𝑙𝑙 𝑘𝑘 𝑠𝑠 + 𝐴𝐴 (A.1)
ows: 𝜅𝜅 𝜅𝜅
shis
OSED study,
added thesubtracted:
ROUGHNESS
and roughness FUNCTION parameters
1
measured by Schultz and Flack [29] are utilised to develop
1 MODEL
+
𝑈𝑈 +model. 1
= 𝑙𝑙𝑙𝑙 𝑦𝑦The + 1 𝑙𝑙𝑙𝑙 𝑘𝑘𝑠𝑠++ + 𝐴𝐴profile on a rough flat plate can be (A.1)
w roughness
B is added function
and subtracted: 𝜅𝜅 + − 𝜅𝜅 velocity expressed as
(A.1)
+ 𝑈𝑈 1 = + 𝑙𝑙𝑙𝑙 𝑦𝑦 − 1 𝑙𝑙𝑙𝑙 𝑘𝑘 + + 𝐴𝐴
𝑠𝑠 + 𝐵𝐵 − 𝐴𝐴)
ows: 𝑈𝑈 = 𝑙𝑙𝑙𝑙 𝑦𝑦𝜅𝜅 + 𝐵𝐵 − ( 𝜅𝜅 𝑙𝑙𝑙𝑙 𝑘𝑘 𝑠𝑠 (A.2)
study,
addedtheandroughness
subtracted: parameters
𝜅𝜅 measured𝜅𝜅 by Schultz and Flack [29] are utilised to develop
added and subtracted:
roughness function model. The velocity profile on a rough flat plate can be expressed as
𝑦𝑦 +1+ + ΔU −−( 11can
1 +1
comparing Eqs. (2) 𝑈𝑈 ++ = and 1+ 𝑙𝑙𝑙𝑙(A.2), 𝐵𝐵 + 𝑙𝑙𝑙𝑙 𝑘𝑘be+ written as:
𝑠𝑠+𝑠𝑠++ 𝐵𝐵 − 𝐴𝐴) (A.1)
(A.2)
𝑈𝑈 = 𝑙𝑙𝑙𝑙 𝑦𝑦𝜅𝜅 + 𝐵𝐵 − ( 𝜅𝜅 𝑙𝑙𝑙𝑙 𝑘𝑘𝑠𝑠 ++𝐵𝐵𝐴𝐴− 𝐴𝐴)
𝑈𝑈 = 𝑙𝑙𝑙𝑙 𝑦𝑦 𝑙𝑙𝑙𝑙 𝑘𝑘 (A.2) Fig. A.1. Rough wall mean velocity profiles [29].
s: 𝜅𝜅
𝜅𝜅
𝜅𝜅
𝜅𝜅
(A.2)
s added By and subtracted: 1 (A.3)
omparing comparing
Eqs. (2)𝑈𝑈and Eqs.(A.2),
+ Δ𝑈𝑈
1(2) = + ΔU
+ and (A.2),
𝑙𝑙𝑙𝑙1+ 𝑘𝑘+𝑠𝑠+can ΔU
+ +𝐵𝐵+be can
− 𝐴𝐴 be written
written as: as: (A.1)
comparing Eqs. (2)+ and1 𝜅𝜅(A.2), = 𝑙𝑙𝑙𝑙 𝑦𝑦 𝜅𝜅 ΔU
−
𝜅𝜅
𝑙𝑙𝑙𝑙 𝑘𝑘 +
can be written as:
1 𝑠𝑠 +
𝐴𝐴 Fig. A.1. Rough wall mean velocity profiles [29].
The roughness constant C is determined by re-calling
s introduced as the = 𝑙𝑙𝑙𝑙 𝑦𝑦 + +
𝑈𝑈 sand-grain 𝐵𝐵 −
roughness ( 𝑙𝑙𝑙𝑙 𝑘𝑘 +
constant, 𝐵𝐵 − 𝐴𝐴) which is defined as follows: (A.2)
dded and subtracted: 𝜅𝜅Δ𝑈𝑈 + = 11 𝑙𝑙𝑙𝑙 𝑘𝑘𝑠𝑠+𝜅𝜅+ 𝐵𝐵 − 𝐴𝐴 𝑠𝑠 Eq. (A.5), as follows: (A.3)
Δ𝑈𝑈 +1= 𝜅𝜅 𝑙𝑙𝑙𝑙 𝑘𝑘𝑠𝑠+ + 𝐵𝐵 − 𝐴𝐴 The roughness (A.3) constant C is determined (A.3) by re-calling Eq. (A.5), as follows:
𝜅𝜅 𝐶𝐶ΔU
𝑙𝑙𝑙𝑙 roughness + (A.4)
𝑠𝑠( =𝑙𝑙𝑙𝑙 𝐵𝐵can 𝑘𝑘−𝑠𝑠+𝐴𝐴+
comparing 1 1
introducedEqs. +(2)sand-grain
as the = and 𝑙𝑙𝑙𝑙 𝑦𝑦(A.2),
+ be 𝐵𝐵written
constant,
− 𝐴𝐴) which as: is defined as follows:
s introduced as𝑈𝑈the sand-grain
𝜅𝜅
+𝜅𝜅 𝐵𝐵 −roughness 𝜅𝜅 constant, which is defined as follows: (A.2) 1 𝜅𝜅 Δ𝑈𝑈 +
C = 𝑘𝑘 + 𝑒𝑒 (A.6)
n, Eq. (A.3) can be reparametrized 1 as follows:
the+11sand-grain (A.3) 𝑠𝑠
Cs is introduced asΔ𝑈𝑈 = 𝐶𝐶𝑠𝑠𝑙𝑙𝑙𝑙= 𝑘𝑘𝐵𝐵 +roughness constant, which
+ (A.4)
mparing Eqs. (2) and (A.2), ΔU
𝑙𝑙𝑙𝑙 +
can
1 𝜅𝜅𝐶𝐶𝑠𝑠 = 𝐵𝐵 1− 𝐴𝐴
𝑠𝑠 − be𝐴𝐴𝐵𝐵written
− 𝐴𝐴 as:
Velocity profiles for different surface (A.4)conditions are shown in Fig. A.1. The
is defined
+ as1follows: + 𝜅𝜅 𝑙𝑙𝑙𝑙 1
s introduced Δ𝑈𝑈 as =the 𝑙𝑙𝑙𝑙 𝑘𝑘
sand-grain 𝑠𝑠 + 𝜅𝜅 𝑙𝑙𝑙𝑙 𝐶𝐶
roughness 𝑠𝑠 = 𝑙𝑙𝑙𝑙 𝐶𝐶 𝑘𝑘 +
constant,
𝑠𝑠 𝑠𝑠 = shift 𝑙𝑙𝑙𝑙 𝐶𝐶𝑘𝑘
which
due
+
𝑠𝑠 tois defined
roughness as follows:
in the fully (A.5)
rough surface
regimeconditions
is 4.6 m/s s = 26, ΔU = 4
+ +
n, Eq. (A.3) can be𝜅𝜅 reparametrized
n, Eq. (A.3) can be reparametrized 1 𝜅𝜅 as𝜅𝜅follows: 𝜅𝜅
as follows: Velocity profiles for different
(A.3) are (k
shown
+ +
re C, the roughness Δ𝑈𝑈constant, = 𝑙𝑙𝑙𝑙 𝑘𝑘is 𝑠𝑠 + 𝐵𝐵 − 𝐴𝐴
dependent onFollowing
the type Eq. of roughness
in Fig. A.1.
(A.6), of the
The surface
downward
the roughness under
velocity
constant Cshift due to 0.2667.
becomes roughness
+ 1 + 1𝜅𝜅 1 1 1 (A.4)
ideration. Δ𝑈𝑈 + = 1 𝑙𝑙𝑙𝑙 𝑘𝑘𝑠𝑠+ + 1 𝑙𝑙𝑙𝑙𝑙𝑙𝑙𝑙
𝐶𝐶 𝐶𝐶=𝑠𝑠 =𝐵𝐵 1 − 𝑙𝑙𝑙𝑙
𝐴𝐴 𝐶𝐶 𝑠𝑠 𝑘𝑘 +
𝑠𝑠 = 1 𝑙𝑙𝑙𝑙 𝐶𝐶𝑘𝑘 +
𝑠𝑠(A.4) in the fully rough (A.5)
regime is 4.6 m/s (k +
= 26, ΔU +
= 4.6, and
ntroduced as Δ𝑈𝑈the=sand-grain
𝜅𝜅 𝑙𝑙𝑙𝑙 𝑘𝑘𝑠𝑠 + 𝜅𝜅 𝑙𝑙𝑙𝑙𝑠𝑠 𝐶𝐶𝑠𝑠 =𝜅𝜅 constant,
𝜅𝜅 roughness 𝑙𝑙𝑙𝑙 𝐶𝐶𝑠𝑠 𝑘𝑘𝑠𝑠+ = 𝜅𝜅which 𝑙𝑙𝑙𝑙 𝐶𝐶𝑘𝑘𝑠𝑠+is defined as follows: (A.5) s
re C, the roughness 𝜅𝜅
constant, is dependent𝜅𝜅 𝜅𝜅 The
𝜅𝜅 final κ
roughness = 0.421). Following
function model Eq.
is (A.6), the
presented roughness
in Eq. constant
(7). C
n, Eq.
re (A.3)
C, the can be reparametrized
roughness 1three flow
constant, is dependent as follows:on the type of roughness of the surface under
onhydraulically
the type of roughness of thea(A.4)surface under
experiments identified regimes: a becomes regime,
smooth 0.2667. transitionally
ideration. 𝑙𝑙𝑙𝑙 𝐶𝐶𝑠𝑠 = 𝐵𝐵 − 𝐴𝐴
Then, Eq. (A.3) 1 can be reparametrized as follows: The final roughness function
(A.5) model is presented in Eq. (7).
𝑘𝑘𝜅𝜅𝑠𝑠+ + regime.
ghideration.
regime, and 1 1 1
Δ𝑈𝑈 +a=fully 𝑙𝑙𝑙𝑙rough 𝑙𝑙𝑙𝑙 𝐶𝐶𝑠𝑠 =𝜅𝜅 𝑙𝑙𝑙𝑙 𝐶𝐶𝑠𝑠 𝑘𝑘𝑠𝑠+ = 𝜅𝜅 𝑙𝑙𝑙𝑙 𝐶𝐶𝑘𝑘𝑠𝑠+
Eq. (A.3) can be reparametrized
𝜅𝜅 𝜅𝜅 as follows:
experiments identified three flow regimes: a hydraulically smooth regime, a transitionally
re C, the roughness
experiments identified constant,three flow is dependent regimes: aonhydraulically the type of smooth roughness of thea surface
regime, under
transitionally
h regime, Δ𝑈𝑈
and
+
=
a1 fully
𝑙𝑙𝑙𝑙 𝑘𝑘 +rough
+
1 regime.
𝑙𝑙𝑙𝑙 𝐶𝐶
1
𝑙𝑙𝑙𝑙 𝐶𝐶 𝑘𝑘 +
=
1
𝑙𝑙𝑙𝑙 𝐶𝐶𝑘𝑘 + (A.5) (A.5)
sideration.
h regime, and a𝜅𝜅 fully𝑠𝑠 rough 𝜅𝜅 regime. 𝑠𝑠 𝜅𝜅= 𝑠𝑠 𝑠𝑠 23𝜅𝜅 𝑠𝑠
C, the roughness constant, is dependent on the type of roughness of the surface under
experiments identified three flow regimes: a hydraulically smooth regime, a transitionally
23 on the type
eration. where C, the roughness constant, is dependent 23
gh regime, and a fully rough regime.
of roughness of the surface under consideration.
xperimentsThe identified
experiments three flow identified regimes: threea hydraulicallyf low regimes: smooth regime, a transitionally
regime, aand hydraulically
a fully rough smoothregime. regime, a transitionally 23 rough regime,
and a fully rough regime.
In the rough regime, for a roughness different from the 24
sand-grain, the roughness constant is23different. Hence,
Eq. (A.5) is applicable to any rough surface, while C is
individually determined.

10.2478/pomr-2023-0060
TRANSFER FUNCTION FOR A CONTROLLABLE PITCH PROPELLER

WITH ADDED WATER MASS
Volodimir Leshchev
Igor Maslov
Oleksandr Palagin
Andrii Naydyonov
Department of Ship Power Plants and Systems, Danube Institute of the National University „Odessa
Maritime Academy”, Ukraine
* Corresponding author: v.leshchev@ukr.net (V. Leshchev)
Abstract
The relevance of this study lies in the fact that it presents a mathematical model of the dynamics of the propulsion
system of a ship that takes into consideration the mass of water added to it. The influence of this phenomenon on the
resonant frequencies of the propeller shaft is examined, and a transfer function for a controllable-pitch propeller is
obtained for various operating modes. The purpose of the study is to improve the calculation of the dynamic operating
modes of a controllable-pitch propeller by examining the features of a visual models. The VisSim software package is
used in the study. A visual model is developed that considers the influence of the rotational speed on the value of the
rotational inertia attached to the variable-pitch screw of the mass of water, and a special transfer function is proposed.
The study shows that a transfer function of this type has a loop enabling negative feedback. An analysis of the operation
of the propeller shaft at its resonant frequency is conducted based on the application of frequency characteristics using
the transfer functions obtained. We show that in the low-frequency region, a consideration of the added rotational
inertia using the proposed transfer function leads to a significant difference compared to the result obtained with the
existing calculation method.
Keywords: propulsion system, torsional vibrations, shaft line, added water mass, transfer function, propeller screw
INTRODUCTION process, as reported in [1-5].

One of the existing ways to increase the accuracy of the
When calculating the free oscillations that occur in the calculated results compared to those obtained in practice is
propeller shaft of a ship, the inertia moment of the controllable to search for a solution using a mathematical model based
pitch propeller (CPP) is usually increased by 20–40% to obtain on the construction of a visual image. This issue has been
a match between the calculated results and the experimental widely covered in a great number of studies. For example,
ones. Such a solution is not always optimal, since there is a in [6-10], the problems associated with the hydrodynamic
discrepancy between the specified calculated frequencies and interaction between a propeller and water are investigated,
the actual ones, especially at low angular speeds of rotation. and the authors explore how the unevenness of the water flow
This means that when calculating the dynamic modes of affects the results of the axial and tangential forces acting
operation of the propulsion system of a ship, it is necessary on the propeller. However, it is not shown how the resulting
to use techniques based on a simplification of the physical harmonic components of the moments of these forces affect

the resonant phenomena in the propulsion complex. In
[11-14], the interaction between the propeller and the
elements of the ship’s hull is investigated, and it is shown
that modern computational fluid dynamics solvers can
give correct calculation results when the rudder operates
in the propeller flow, although these studies rarely
consider the features of the dynamics of the propulsion
system from the action of the displaced water mass.
The studies in [15-20] explore the dynamics of a ship’s
Figure 1. Model of the three-mass mechanical part of the propulsion system
propeller when the water surface is agitated using a model that (Md – torque produced by the diesel engine, Mc – resistance torque on the
includes an increase in the rotational inertia of the propeller propeller shaft, Jd, Jg, Jsp – rotational inertia of the engine, gearbox, and
due to the simple addition of the water mass. The difficulty propeller respectively, Cis – stiffness of the intermediate shaft, Cps – stiffness
of the propeller shaft, bis, bps – viscous friction coefficients for the shafts, s –
of this approach means that such studies are limited. Thus, differentiation operator)
despite abundant research in the field of dynamics on the
propulsion system of ships, the issue of the features of stability
and resonant frequency phenomena in the interaction of the Table 1 shows the designations of the model elements used
propeller and the aquatic environment remains unresolved. in the study and their given values. The angular velocities of
The purpose of this study is to improve the calculation of the rotating elements of the propulsion system, after being
the dynamic operating modes of a controllable-pitch propeller connected to the motor shaft, are assumed to be equal:
by examining the features of a visual model.
(1)
MATERIALS AND METHODS All values of frequencies and angular velocities in the text
and figures are given in rad∙s-1 due to the limitations of the
This study involves mathematical modelling of the process VisSim software product.
that occurs during the movement of the ship, that is, the Table 1. Designations and technical parameters of the ship’s propulsion system
interaction between a controllable-pitch propeller and the
aquatic environment for a propulsion system represented
Elements of the
in the form of a three-mass mechanical system (TMS) [21]. propulsion system Symbol
Parameter values Parameter values
In this approach, three distinct masses represent integral before casting after casting
of the ship
components of the propulsion pathway. The first mass Diesel rotational
Jd 4.50·103 kg. m2 4.50kg. m2
corresponds to the diesel engine, and includes its rotational inertia
inertia and the torque it produces. The intermediary mass Rotational inertia
Jr 2.40·103 kg. m2 0.24kg. m2
represents the gearbox or reduction gear, and accounts for of the gearbox
its rotational inertia and its role in moderating the engine Estimated
speed for the propeller. The final mass represents the propeller rotational inertia Jsp 3.2·103 kg. m2 0.32kg. m2
and its shaft, and includes the propeller’s rotational inertia of the screw
and the resistance it faces due to water and other factors. Rotational inertia Jv=(0.2-
(0.6-1.2)·103 kg. m2 (0.6-0.12)kg. m2
The VisSim software package is used here to visualise the of the added water 0.4)Jsp
results of the study. VisSim is a visual-based block diagram Stiffness
programming language tailored for simulation and embedded of the intermediate Cis 0.64·106 N.m 0.64N.m
system development. It has been extensively employed in the shaft
design of control systems and digital signal processing, and Stiffness
facilitates multi-domain design and simulation. The language of the propeller C ps 0.24×106 N.m 0.024N.m
shaft
offers blocks for arithmetic, Boolean, and transcendental
Gear ratio
operations, in addition to digital filters, transfer functions, ig 3.19 3.19
of the gearbox
numerical integration, and interactive graph plotting [22].
Nominal angular
Figure 1 shows a model diagram of the TMS of a ship’s
rotation speed ωdn 40.317rad∙s-1 1
propulsion system [17, 22], with a modified transmission of the diesel engine
function for the CPP in accordance with the one proposed
in this study. Since the action of the dissipative forces of the viscous
action is assumed to be equal to zero, the values of the resonant
frequencies in the shaft line obtained from experiment are
close to the calculated values [22]. The study was conducted
using the VisSim experimental setup shown in Figure 2.
In addition to the TMS, the model shown in Figure 2 also
contains an exciting scanning signal generator (SSG) [17].

Figure 2. Diagram of the dynamic model of the three-mass mechanical part of the propulsion system of the ship
This generator supplies the diesel output with a sinusoidal

signal with a constant amplitude and frequency, which varies (4)
in proportion to time with acceleration .This
signal simulates the rotation of the diesel engine, which leads
to resonant phenomena in the shaft line. where n is the speed of rotation of the propeller, and ωsp is its
The operation of the model and the interaction of its rotational speed (rad/s).
parts are described in [22]. Fixed frequency generators G1 If the rotating mass of water is a full-bodied cylinder with
and G2 are designed to check and refine the values found radius:
during scanning of the critical frequencies Ωcr1 and Ωcr2. The
experimental equipment for the study was provided by the (5)
Danube Institute of the Odessa Maritime Academy.
then, from Eq. (4), the following can be obtained:
RESULTS AND DISCUSSION (6)
It is known that in a ship’s propulsion system, a fixed-

pitch propeller (FPP) under ideal conditions, taking into In the existing method of dynamic calculation, the problem
account the slippage of the propeller blades relative to the of finding the moment:
water, displaces and swirls a unit mass of attached water per
one revolution: (7)
(2) is reduced to simply increasing it by 20–40% of the calculated

value:
where mvE is the unit mass of the added water formed from
one turn of the screw; Hsp is the pitch of the propeller; Dsp is (8)
the diameter of the propeller; ρ is the density of water; and
α is the coefficient of sliding. This allows for some coincidence between the results of
When using the above units, the angular frequency of a the calculation of critical frequencies with their real values.
rotation and the rotational speed are related by the expression: However, it follows from Eq. (6) that the moment Jv is not
constant, but changes over time depending on the changes
(3) of the variable components ωsp(t) and Hsp(t). The dynamics
equation for the screw in this case takes the form:
The mass of the added water jet mv at the rotational speed
of the propeller ωsp(rad/s) can then be represented by the
formula:

(16)
where is the torque from

the water added to the propeller.
If we assume that the moment of the load Mc(s) = 0 then
under zero initial conditions of the dynamic process, Eq. (16)
in the operator form is:
(9) (17)
The second part of Eq. (9) represents the torque of the

displaced water mass Mv, which contains the differential of where s is the Laplace operator.
the product of the variables H(t) and ωsp(t). This product is Since the propeller is affected by two torques, Mps(s) and
differentiated as follows: Mv(s), the mathematical model of this interaction takes the
form shown in Figure 3.
(10)
If the expression in Eq. (10) is substituted into Eq. (9),

then we have:
(11)
The terms of this equation by derivatives after grouping are:
Figure 3. Diagram of the mathematical model of the dynamic interaction

(12) between torque and a unit moment
In the diagram of the propulsion system model in Figure

This expression for the moment Mps holds both for a unit 2, this node is highlighted in colour. It contains the values
with an FPP and for a unit with a CPP. It describes a single of the elements that correspond to numerical experiment.
dynamic process that occurs when both accelerating or As can be seen from Figure 3, the moments Jsp and JvE are
decelerating ωps(t), and when changing the step Hsp(t) towards combined into a loop containing negative feedback. The total
an increase or decrease. There is also a scenario where these transfer function of the FPP can be then presented in
two cases act simultaneously. In this study, we investigate the generalised form as:
option in which the ship has FPP or a CPP that operates with
a given fixed step Hsp. Eq. (12) then takes the form:
(18)
(13)
where Mps is the torque of the propeller shaft with the FPP.
Part of the second term in Eq. (13) can be interpreted as It follows from Eq. (18) that the model of the total transfer
the rotational inertia of the added water for one revolution function of the propeller has the form shown in
of the propeller, and is expressed by the formula: Figure 4.
(14)
Considering the damping effect of water, we can write Eq.

(13) with respect to the rotational inertia of the propeller Jsp
as follows: Figure 4. Model of the generalised transfer function of the CPP
considering the added water mass
(15)
Then, the equation for the propeller dynamics takes the The transfer function of the FPP found here differs
form: significantly from the one used today in practice, which is
accepted as:

(25)
(19)
The transfer function Wv(s) is implemented as a typical
It can also be seen from Eq. (18) that the value JvE is specific differentiating link:
to each case, and is subject to calculation. The moment Jv is
inversely proportional to the unit moment JvE. It follows from (26)
this that with a decrease in the step ratio
The total value of the moment JΣ in this case is equal to:
(20)
(27)
the critical frequency Ωcr1 shifts downwards. For a TMS
propulsion system, this displacement mainly affects low
rotational speeds, and may go beyond the range that After modelling the dynamic process, the critical
corresponds to (1.2 - 1.4) Jsp. When calculating the dynamic frequencies are found by scanning the model with the SSG
modes, attention should be paid to this phenomenon. A variable signal generator for the three variants, and have the
numerical experiment is conducted with a dynamic model following values: Ωcr1-1 = 0.2081rad*s-1, Ωcr1-2 = 0.22329rad*s-1,
as shown in Figure 2, where the values of the elements of Ωcr1-3 = 0.2603rad*s-1, Ωcr2-1 = Ωcr2-2 = Ωcr2-3 = 1.4332rad*s-1.
this model are shown in the diagram in accordance
with Table 1. Descriptions of the interaction between
the parts of the model are presented in [23-26]. For
example, the propulsion system of the ship is taken as
an FPP with the following technical characteristics:
step ratio λsp = 1.045; coefficient of relative slip α =
0.14; screw diameter Dsp = 1.1 m; and engine 6NVD
48A-2U with rated power 852 kW (1158.4 hp), rated
engine speed ndn = 385 rpm, and rated speed of
the propeller shaft nsp = 118 rpm. A comparative
analysis of the dynamics of the propulsion system is
conducted for three scenarios. In the first, the FPP
has an estimated rotational inertia of
(21)
Figure 5. Critical frequencies of the torque Ωcr1-1, Ωcr1-2 , and Ωcr1-3 for the
and in the second, the FPP has a rotational inertia that is TMS propulsion system for various values of the moment Jv
increased by 40%:
(22) As can be observed from the oscillogram in Figure 5,

the high critical frequencies of the torque , Ωcr2-1, Ωcr2-2 and
The third involves a screw with a rotational inertia Ωcr2-3 are almost the same. This allows us to conclude that
determined using the method proposed in this study in the rotational inertia of displaced water Jv has little effect on
accordance with Eq. (18). The unit moment JvE is calculated critical frequencies that occur below the maximum engine
by Eq. (2), where the screw pitch Hsp is replaced by a step ratio speed [27-31].
λsp. Then, for ωspn = 1, Eq. (2) becomes:
(23) CONCLUSION
where is the moment reduction coefficient. Our results show that for the propulsion system considered
here, the critical frequency Ωcr1-2 obtained using the propeller
By substituting the values of the circuit elements (Table 1) transfer function proposed in this study is within the existing
into Eq. (23), the following estimate can be obtained: range of changes in critical frequencies, which are typically
estimated in practice by increasing the moment Jsp by 20–40%.
(24) It corresponds to a total moment Jsp = 0.3885 kg ∙ m2, which
represents an increase of 21.9% and confirms the validity of
its application. The proposed method for determining the
The transfer function of the added moment Wv then takes rotational inertia of a screw with a displaced water mass
the value: allows us to calculate the ratio directly, without the need for
approximate empirical formulae such as Kutuzov’s formula,

which does not consider the dependence of the moment Jv on with a cracked functionally graded shaft − finite element
the rotation frequency of the mass of the water jet. analysis. European Journal of Mechanics − A/Solids, 61,
A visual dynamic model was developed that considered the 47-58.
influence of the rotational speed on the value of the rotational
inertia of water mass added to the CPP, and a special transfer 7. Lou, B., Cui, H. 2021. Fluid–structure interaction vibration
function for the propeller was proposed. It was established experiments and numerical verification of a real marine
that such a transfer function has a loop with negative feedback. propeller. Polish Maritime Research, 28(3), 61-75. https://
The operation of the ship’s shaft line when using the proposed doi.org/10.2478/pomr-2021-0034
transfer functions in the model of the propulsion system was
explored. An analysis of the resonant modes based on the 8. Prohl, M.A. 1945. A general method for calculating critical
application of frequency characteristics using the transfer speeds of flexible rotors. Journal of Applied Mechanics,
functions obtained in the study was conducted. It was found 12(3), 142-148.
that when determining the critical oscillation frequencies
of the shaft line in the lower part of the propeller operating 9. Ostanin, V. 2022. Vadym Effects of repulsion and attraction
range, considering the added moment leads to a significant between rotating cylinders in fluids. Scientific Herald of
difference compared to the results obtained from the existing Uzhhorod University. Series “Physics”, (51), 39-47. https://
method. doi.org/10.54919/2415-8038.2022.51.39-47
NOMENCLATURE 10. Klendii, M., Logusch, I., Dragan, A., Tsvartazkii, I.,
Grabar, A. 2022. Justification and calculation of design
and strength parameters of screw loaders. Machinery
CPP Controllable-pitch propeller & Energetics, 13(4), 48-59. https://doi.org/10.31548/
machenergy.13(4).2022.48-59
FPP Fixed-pitch propeller
11. Yoon, M. 2016. A Transfer Function Model of Thrust
SSG Scanning signal generator Dynamics for Multi-Rotor Helicopters. International
Journal of Engineering Research & Technology, 5(1), 15-18.
TMS Three-mass mechanical system
12. Boletis, E., de Lange, R., Bulten, N. 2015. Impact of
propulsion system integration and controls on the vessel
REFERENCES DP and maneuvering capability. IFAC-PapersOnLine,
48(16), 160-165.
1. MacPherson, D.M., Puleo, V.R., Packard, M.B. 2007.
Estimation of entrained water added mass properties for 13. Xiros, N.I. 2004. PID marine engine speed regulation
vibration analysis. https://cutt.ly/8GVZdOe. under full load conditions for sensitivity H∞-norm
specifications against propeller disturbance. Journal of
2. Faltinsen, O., Minsaas, K.J., Liapias, N., Skjørdal, S.O. Marine Engineering & Technology, 3(2), 3-11.
1981. Prediction of resistance and propulsion of a ship in
a seaway. In: T. Inui (Ed.). Proceedings of 13th Symposium 14. Smailova, G., Yussupova, S., Uderbaeva, A., Kurmangaliyeva,
on Naval Hydrodynamics (pp. 1-19). Tokio: University of L., Balbayev, G., Zhauyt, A. 2018. Calculation and
Trondheim. construction of the tolling roller table. Vibroengineering
Procedia, 18, 14-19. https://doi.org/10.21595/vp.2018.19908
3. Król, P. 2021. Hydrodynamic state of art review: Rotor
– stator marine propulsor systems design. Polish 15. Koushan, K. 2006. Dynamics of ventilated propeller blade
Maritime Research, 28(1), 72-82. https://doi.org/10.2478/ loading on thrusters. In: World Maritime Technology
pomr-2021-0007 Conference (pp. 18-21). London: Macmillan Education.
4. Koibakov, S.M., Umirkhanov, M.G. 2013. Model research 16. Senjanović, I., Hadžić, N., Murawski, L., Vladimir, N. 2019.
of ice jams. World Applied Sciences Journal, 25(8), 1158- Analytical procedures for torsional vibration analysis of
1160. https://doi.org/10.5829/idosi.wasj.2013.25.08.13382 ship power transmission system. Engineering Structures,
178, 227-244.
5. Koibakov, S.M., Umirkhanov, M.G. 2013. Icebreaker unit.
World Applied Sciences Journal, 25(8), 1251-1254. https:// 17. Leschev, V.A. 2018. Marine diesel ACS with external
doi.org/10.5829/idosi.wasj.2013.25.08.13423 feedback speed sensor. Modern Engineering and Innovative
Technologies, 5, 11-17.
6. Gayen, D., Chakraborty, D., Tiwari, R. 2017. Whirl
frequencies and critical speeds of a rotor-bearing system

18. Kukhar, V., Vasylevskyi, O., Khliestova, O., Berestovoi, I., 29. Xiang, L., Yang, S.X., Gan, C.B. 2012. Torsional vibration
Balalayeva, E. 2022. Hydraulic Press Open Die Forging of a shafting system under electrical disturbances. Shock
of 21CrMoV5-7 Steel CCM Roller with Flat Upper and and Vibration, 19, 1-11.
Concave Semi-round Lower Cogging Dies. Lecture
Notes in Mechanical Engineering, 489-498. https://doi. 30. Quang, P.K., Hung, P.V., Cong, N.C., Tung, T.X. 2022. Effects
org/10.1007/978-3-030-91327-4_48 of rudder and blade pitch on hydrodynamic performance
of marine propeller using CFD. Polish Maritime Research,
19. Aghbalyan, S., Simonyan, V. 2022. Study of hardening and 29(2), 55-63. https://doi.org/10.2478/pomr-2022-0017
structure of maraging powder steel grade PS-H18K9M5TR
(18%Ni+9%Co+5%Mo+1%Ti+1%Re+66%Fe). Scientific 31. Fleischer, K.P. 1973. Untersuchungen über das
Herald of Uzhhorod University. Series “Physics”, (52), Zusammenwirken von Schiff und Propeller bei
46-55. https://doi.org/10.54919/2415-8038.2022.52.46-55 teilgetauchten Propellern. Forschungszentrum des
Deutschen Schiff baus Bericht, 130(73), 291-308. [in
20. Nussupbek, Z.T., Bekenov, T.N., Sattinova, Z.K., Beisenbi, German].
M.A., Tassybekov, Z.T. 2023. Substantiation of methods
for calculation of traction forces redistribution indicators
on modular front and rear wheels of the vehicle (4Х4).
Transportation Engineering, 13, 100193. https://doi.
org/10.1016/j.treng.2023.100193
21. Gierusz, W. 2016. Modelling the dynamics of ships with

different propulsion for control purpose. Polish Maritime
Research, 89(23), 31-36.
22. Guimarães, D. A. 2009. Digital Transmission: A Simulation-

Aided Introduction with VisSim/Comm. New York:
Springer Verlag. https://doi.org/10.1007/978-3-642-01359-1
23. Leshchev, V.A., Naydyonov, А.I. 2021. Dynamic Method for

Determining Resonant Frequencies of Torsional Vibrations
of a Ship’s Propeller Shaft. A Scientific Look into the Future,
1(21), 15-26.
24. Gorb, S., Popovskii, A., Budurov, M. 2023. Adjustment

of speed governor for marine diesel generator engine.
International Journal of GEOMATE, 25(109), 125-132.
https://doi.org/10.21660/2023.109.m2312
25. Califano, A. 2010. Dynamic loads on marine propellers due

to intermittent ventilation. Trondheim: NTNU.
26. Kaplun, V., Chuenko, R., Makarevych, S. 2022. Investigation

of energy parameters of a compensated asynchronous
motor in the mode of repeated short-term starts. Machinery
& Energetics, 13(3), 25-33. https://doi.org/10.31548/
machenergy.13(3).2022.25-33
27. Ghaemi, M.H., Zeraatgar, H. 2022. Impact of propeller

emergence on hull, propeller, engine, and fuel
consumption performance in regular head waves. Polish
Maritime Research, 29(4), 56-76. https://doi.org/10.2478/
pomr-2022-0044
28. Kluczyk, M., Grządziela, A., Batur, T. 2022. Design and

operational diagnostics of marine propellers made of
polymer materials. Polish Maritime Research, 29(4), 115-
122. https://doi.org/10.2478/pomr-2022-0049

10.2478/pomr-2023-0061
Digital Twin test-bench Performance

for marine diesel engine applications
Dmytro Minchev 1
Roman Varbanets 1
Oleksandr Shumylo 1
Vitalii Zalozh 2
Nadiia Aleksandrovska 1
Pavlo Bratchenko 1
Thanh Hai Truong 3

1
Odesa National Maritime University, Odesa, Ukraine
2
Danube Institute of the National University „Odesa Maritime Academy”, Ukraine
3
DPATET Research Group, Ho Chi Minh City University of Transport, Ho Chi Minh City, Viet Nam
* Corresponding author: dmytro.minchev@nuos.edu.ua (D. Minchev)
Abstract
The application of Digital Twins is a promising solution for enhancing the efficiency of marine power plant operation,
particularly their important components – marine internal combustion engines (ICE). This work presents the concept
of applying a Performance Digital Twin for monitoring the technical condition and diagnosing malfunctions of marine
ICE, along with its implementation on an experimental test-bench, based on a marine diesel-generator. The main
principles of implementing this concept involve data transmission technologies, from the sensors installed on the engine
to a server. The Digital Twin, also operating on the server, is used to automatically process the acquired experimental
data, accumulate statistics, determine the current technical state of the engine, identify possible malfunctions, and make
decisions regarding changes in operating programs. The core element of the Digital Twin is a mathematical model of the
marine diesel engine’s operating cycle. In its development, significant attention was devoted to refining the fuel combustion
model, as the combustion processes significantly impact both the engine’s fuel efficiency and the level of toxic emissions of
exhaust gases. The enhanced model differs from the base model, by considering the variable value of the average droplets’
diameter during fuel injection. This influence on fuel vapourisation, combustion, and the formation of toxic components
is substantial, as shown. Using the example of calibrating the model to the test results of a diesel engine under 27 operating
modes, it is demonstrated that the application of the improved combustion model allows better adjustment of the Digital
Twin to experimental data, thus achieving a more accurate correspondence to a real engine.
Keywords: digital twin, combustion model, marine diesel engine, diagnostics
INTRODUCTION as the global market for Digital Twin products was valued
at USD $3.8 billion in 2019 and is expected to reach USD
Digital twin technology is a promising solution for current $35.8 billion by 2025 [2].
and future marine power plants. It could be used in various The concept of Digital Twins has been experiencing rapid
ways for various purposes: advanced control, monitoring and development in recent years. Today, Digital Twins are classified
diagnostics, management and data analysis. as the following types: Digital Twin Instance (DTI), Digital Twin
The idea of Digital Twin technology was originally Prototype (DTP) and Performance Digital Twin (PDT) [3].
developed for product lifecycle management issues, by M. However, all types of Digital Twins have common characteristics:
Grieves in 2002, and subsequently refined by J. Vickers of high-fidelity, dynamic, self-evolving, identifiable, multiscale,
NASA in 2010 [1]. It proved to be a very promising approach, metaphysical, and hierarchical [4], [5].

According to some specialists, the Digital Twin could mixing with air and combustion are extremely complicated.
be described as having the highest possible bidirectional Although some researchers reported the application of
integration level between the physical object and virtual model. simplified combustion models, based on the Wiebe function
In contrast, the Digital Shadow only reflects the state of the [15], it has a huge limitation, in terms of predicting fuel timing
physical object while the Digital Model has no automatic and atomisation parameters for combustion processes.
interactions with the physical object [4]. Major areas for the This paper presents developments in the proven Razleitsev
implementation of Digital Twins are: smart cities and urban combustion model for marine diesel engines with Digital
centers, freight logistics, medicine, engineering and automotive Twins and the research test-bench, based on the marine diesel-
technologies. generator, which proves the concept. The research focuses
A major area of Digital Twins applications is fault diagnostics on improvements in fuel evaporation and mixing prediction
and continuous monitoring of an object’s tecnnical state. By and its effect on further combustion and pollutant formation
combining Digital Twins with the deep transfer learning processes.
approach, accurate machine fault diagnosis could be provided,
even with insufficient measured fault condition data [6]. The
Digital Twin is continuously updated to generate possible DigitAL Twin for engine
fault conditions close to the actual asset and constructs the monitoring and control
training data in the source domain for transfer learning. Digital
Twin technology also can provide the visualisation of the A Digital Twin type of marine diesel engine serves the
monitoring process for the object being monitored, including following purposes: monitoring of an engine’s technical state,
3D-visualisation and augmented reality technologies [7]. The diagnostics of possible malfunctions, and control of the engine
very promising predictive maintenance method, based on operating mode. For these purposes, the Digital Twin has to
Digital Twin, could be applicable in many areas and presents include the mathematical model of the engine operating cycle,
three unique characteristics: real-time perception, a high which meets the requirements of accuracy and computational
fidelity model and high confidence simulation prediction [8]. speed.
For a marine power plant and its components, Digital Twin The diagram of the Digital Twin-based engine monitoring
could serve various purposes, among which are: prototyping system is shown in Fig. 1. The engine is equipped with an array
and design, technical state monitoring and diagnostics, control of sensors, allowing measurement of the engine’s operating
and efficiency analysis, lifecycle management. It could also be parameters. All of the sensors transfer the data to the set of
a solution for environmental impact monitoring from shipping, controllers, which transmits the data to a Web Server via
which is one of the IMO goals with high priority [9]. Various a Wi-Fi wireless connection.
components of marine power plants require monitoring [10], The Web Server contains the Database, which stores the
[11], while diesel engines remain one of the most important. measured data, as well as the data generated by the Digital
In terms of marine diesel engines, the most challenging Twin. The Digital Twin autonomously takes data from the
factors for the successful implementation of Performance Database and runs the necessary calculations to compare
Digital Twin are: accurate engine operation simulation and the measured data with the predicted engine performance.
enabling fast calculations. Obviously, for engine control tasks, The visualisation of the actual engine operation and the
it is necessary to provide ‘real-time’ engine operating process predicted engine performance is provided in a pseudo real-
calculations [12], [13]. The ultimate goal should be defined time mode for educational and scientific purposes. Based on
as: engine operating cycle synthesis time equal to the actual the comparison of the experimental and predicted engine
engine operating cycle time. It is always a trade-off between performance, the Digital Twin provides diagnostic information
calculation time and accuracy, which should be considered related to the conditions of engine subsystems, such as: the
at the decision making stage of choosing an appropriate turbocharging system, the fuel injection system and the gas
mathematical model for the Digital Twin core. distribution mechanism.
Engine manufacturers can potentially adopt high-fidelity The Digital Twin also has to set up the user interface before
Digital Twins for real-time engine control issues, as they have it can be used in an actual engine.
all the complete information about engine sub-systems and
components. The heuristic-type model could be applied in
this case [14]. In contrast, the monitoring systems with Digital
Twins provided by external independent developers should
have the characteristic of flexible tuning and adaptation, even
in the absence of data.
One of the most critical things for the sufficient operation of
Digital Twins is the correct prediction of the heat release process
in the engine cylinder due to the fuel burning. Simulation of
fuel combustion represents the toughest challenge, in terms
of the development of the Digital Twin mathematical model,
as the processes of fuel injection, atomisation, evaporation, Fig. 1. Digital Twin application for engine control and monitoring issues.

Fig. 2. Test-bench measuring points diagram:1 – engine; 2 – AC alternator; 3 – exhaust manifold; 4 – intake receiver; 5 – turbine; 6 – compressor;
7 – exhaust silencer; 8 – air mass flow meter; 9 – fuel mass flow meter; 10 – high-speed analogue-to-digital converters
and transmitters; 11 – set of slow-speed analogueue-to-digital converters and transmitters.
We considered some detailed peculiarities of the concept

using the actual test-bench as an example. The test-bench is
based on the Weichai WP4C82-15 marine diesel-generator
and was installed in the Odesa National Maritime University
Laboratory.
The measuring device diagram is presented in Fig. 2 and it
shows the measuring points and parameters. The measuring
system can be considered as consisting of two sub-systems:
‘static’ and ‘dynamic’. The former has low-speed parameters,
which are measured as averages of the number of engine
crankshaft revolutions, and the latter are high-speed parameters,
which show the engine operating cycle as a function of the crank
angle degree (the measuring step is 0.5-1.0 c.a.d.).
Static parameters include intake and exhaust system
temperatures and pressures, engine power, air and fuel flow.
The turbocharger speed is estimated by an acoustic method, as
Fig. 3. Database structure.
described in [16]. Dynamic parameters embrace in-cylinder
pressure, pressures at the exit of the fuel injection pipe and Among other things, these maps allow for the consideration of
before the fuel injector and the vibration sensor signal. The phenomena such as compressor surge in a turbocharger [21].
vibration sensor provides vibroacoustic data concerning fuel The mathematical model of fuel combustion is based on
injection, and intake and exhaust valve timings [17], [18]. the Razleitsev method for the equivalent fuel spray approach
The obtained data is preprocessed using a special treatment and the toxic emission estimation implements Zvonov’s
for indicated diagrams analysis [19]. method of burned zone gas composition prediction. As the
The database diagram, shown in Fig. 3, shows its basic fuel combustion processes have a significant impact on the
structure and includes tables for both static and dynamic performance parameters of the engine and the level of toxic
parameters. emmisions in the exhaust gases, as well as the fact that these
The Digital Twin is based on the Blitz-PRO online service, processes have an inherently complex physicochemical nature,
providing simulations of both the static and transient operation one of the most important components for the mathematical
of Internal Combustion Engines of various types [20]. The model of a Digital Twin is a precise fuel combustion model.
mathematical model of the engine’s operating cycle is based As the fuel combustion depends greatly on the movement
on the two-zone quasi-steady approach for in-cylinder process of fuel spray, the injection rate diagrams, fuel atomisation
synthesis, one-dimensional simulation of the intake and exhaust and in-cylinder air motion, the application of a complex
pipes processes, and advanced supercharger performance map combustion model is challenging; this type of information is
treatment. hardly achievable at engine operating conditions. From this
The important task of providing the closed-loop modelling of point of view, the combustion model should be simple enough
the operating process of a turbocharged engine is solved using to set up and apply to practical power plants and accurate
the advanced characteristic maps of supercharging devices. enough to correctly replicate engine performance.

Improvements in fuel is the reduction in the evaporation rate which occurs when
combustion model fuel interacts with the walls.
It should be noted that Razletsev and his followers, instead of
The fuel combustion model for diesel engines was proposed using the accurate solution mentioned above, used a simplified
for the equivalent spray approach by Razletsev in 1980 [22], and equation, assuming a constant injection rate over the injection
further developed in 1992 [23], to account for the peculiarities process time and equal to the average value:
of fuel jet motion with a detailed mechanism of interaction
dσev 1 χ([1– bev (φ – φin j)]3/2–[1– bev]3/2).
between the fuel flame and the combustion chamber walls. dφ = φin j (4)
Subsequently, Razletsev’s model was slightly improved by
Kuleshov [24] and implemented in commercial software tools The use of the simplified equation lead to a significant error
for the calculation of engine operating processes. in the calculation results, as shown in Fig. 4. For nonlinear
Razletsev’s method involves a sequential and interconnected injector rate profiles, the cumulative error of the evaporated
consideration of the fuel injection, evaporation, mixture fuel share could exceed 20% and so the application of the
formation, and fuel combustion using fundamental laws of accurate solution is important.
physics and chemical kinetics.
The characteristics of fuel injection are determined using
experimental or statistical data. The basic Razletsev method
assumes single injection events but it can be adopted for
multiple injections, as we will show later. The movement of
fuel sprays and their atomisation is determined using criteria
equations proposed by Lyishevsky.
τ
la = 1.22 · lc · ρ–0.5Эд–0.35 · Э0.35e–0.2( τ ) ; c
– – 0.5
lb = din j.holesu0We0.21M 0.16 / (3√ 2ρ )τ ;
––0.6 ;
lc = 8.85din j.holesWe0.25M 0.4 ρ
(1)
d32 = 106Ecdin j.holes (ρ–M
0.0733
We)0.266 ; Fig. 4. Comparison of fuel injection rate diagrams calculated

μ2 (2)
– = ρc ;
M = din j.holes ρfuelfuel σfuel ; ρ
using simplified and accurate equations
ρfuel
τ2σfuel The fuel evaporation constant is dependent on the average
u d
2
ρfuel
We = in j inσj.holes
fuel
; Э = d in j.holesρfuel
3 ; diameter of the fuel spray droplets d32 and can be calculated
using the following expression:
where la, lb and lc are the spray length development at the initial
bev = Y 10 1
12
stage, the main stage of injection and at the end of the initial d2 pc , (5)
32
stage; τ and τc are time moments from injection start and from
injection start to the end of the initial stage of injection; μfuel, where Y is the correction factor; pc is the end pressure of the
ρfuel, and σfuel are the dynamic viscosity, specific gravity and compression process; and mY is the exponent of the evaporation
surface tension of the fuel; Ec is a coefficient; and din j.holes is the function.
nozzle hole diameter. According to Razletsev’s recommendations, the correction
Fuel atomisation is characterised with the Sauter diameter factor Y can be calculated using the following expression:
of fuel droplets d32, which is the ratio of the total volume of fuel
ncrank
Y = ( 1000 ) mY
drops to their combined surface area. For simplification reasons , (6)
the value d32 was considered to be constant by Razletsev and
equal to the average for the injection period [22], [23]. where mY is the exponent of the evaporation correcting
To calculate the fuel evaporation, a modification of function.
Sreznevsky law for the evaporation of an individual droplet Typically, mY is chosen within the range 0.65-1.00. However,
is used. To determine the fraction of fuel evaporated at a given if the crankshaft rotational speed of the engine is below
moment in time, integration is performed for all fuel portions 1000 rpm, the value of mY can be selected within the range
injected up until that moment: 0.35-0.70. These values allow for appropriate adjustments
in the evaporation function, to account for different engine
dσev φ–φin j.start 0.5 dσ
dφ = ∫ φin j.start 2 wallbev(1– wallbev 6n ) [ dφ ]dφ,
φ 3χ χ operating conditions.
(3) Kuleshov proposed a more complex expression for the
function Y:
where σev is the evaporated portion of the fuel; dσ/dφ is the
relative injection rate; bev is the evaporation constant; and χwall Y = 0.372 · 10–9(18+ys+yrpm)yHy0.35d32–1.5 , (7)

where yS = f(S) is the scaling factor dependent on the piston is possible for a period of Δφk.ext by setting the parameters Δφk
stroke S; yrpm = f(ncrank) is the correction coefficient dependent and Δτk.
on the crankshaft rotational speed ncrank; Hy is the corrected Typically, for the external characteristic, Δφk = 0, reaching
swirl number; and y is an empirical coefficient ranging values of 5-12 in idle modes, while Δτk is recommended to
from 5-35. be selected in the range of 0.3-0.8 for direct injection and
This more comprehensive expression for the function Y 0.5-0.9 for split combustion chambers. The basic system of
considers additional factors, such as piston stroke, crankshaft equations is as follows:
rotational speed, and the corrected swirl number, to provide
a more accurate estimation of the evaporation constant. From
the given expression, we can conclude that the evaporation
constant is variable during fuel injection and depends on the
average diameter of the droplets.
The relative reduction in the evaporation rate when the fuel
sprays come into contact with the combustion chamber walls,
according to the original methodology [22], can be calculated (10)
using the following equation: where σi is the fraction of fuel supplied during the ignition
φ–φwall 2 delay period; φinj.end is the end of injection moment; Δφk.ext is
χwall = 1 – ( 1–χ0 ) · 0.707( φ–φ 2 –0.5( φfr )
φfr ) · – e ;
wall
0.485 the extended period of applying the second equation; Δφcomb.end
√π
is the end of the combustion process; ξa.c is the air utilisation
ρ–0.5We0.32
φfr = Ast · 2φwall M0.07 , (8) function; and ΔU.F is the unburned fuel fraction.
Functions P0, P2, A0, and A2 are determined according to
where φwall is the moment when the spray reaches the the equations:
combustion chamber wall; χ0 is the minimum value of the
A q (σ –x )
reduction coefficient of the evaporation rate; φfr is the duration P0 = V(φ
0 fuel ev 0
(b0 σev + x0) ;
comb.start)
of the interaction between the spray and the combustion
A2qfuel(α–x)
chamber walls; and Ast is the coefficient in the formula for P2 = Vc
(σev – x);
calculating the cone angle of the fuel spray. (11)
To calculate the ignition delay period, a modified Tolstov
equation is used: A0 = a0 (n · H)mcomb ;
p in. j.start RT
E
70a A1 = a1 (n · H)mcomb ;
A2 = a2 (n · H)mcomb ,
in. j.start
τi = B0(1– knncrank) Tcylin. j.start e 25+CN , (9)

cyl

cyl
(12)
in. j.start in. j.start
where B 0 and kn are coefficients; pcyl and T cyl are the where H is a swirl number; and a0, a1, a2, b0, and mcomb are
pressure and temperature at the beginning of compression; adjustable coefficients.
and Ea and CN are the activation energy and cetane number
of the fuel.
In a further development, a more advanced model of the
fuel spray was proposed, which includes the following seven
characteristic zones [23]: dense axial core, dense front jet,
rarefied jet envelopee, conical axial core of the boundary flow,
boundary flow core on the piston surface (head, cylinder
liner), and rarefied boundary flow envelope. For each of these
characteristic zones, the fuel evaporation constant is calculated
using individual equations.
The application of a more advanced model of fuel sprays
complicates the combustion model tuning process and requires
accurate information about the geometry of the combustion
chamber and fuel injector, which is not always available.
During the calculation of heat release, the following processes
are considered separately: combustion of the air-fuel mixture
formed during the ignition delay period, diffusion combustion
in the fuel supply region, and post-injection fuel burning after
the end of injection. The transition between equations for
each stage occurs at specific time points: when x = σi from the
equations for the first stage to the equations for the second stage,
and when φ = φinj.end + Δφk.ext from the equations for the second Fig. 5. The relationship between fuel injection, fuel vapourisation processes,
stage to the equations for the third stage. The continuation of and fuel combustion for the Razleytsev model, along with the model tuning
using the equations for the second stage after the end of injection parameters for a single injection case.

Tab. 1. Recommended values for fuel combustion model
Engine type a 0 · 10 -3 a1 · 102 a2 b0 mcomb H mY
n = 50…250 rpm, two-stroke 5…12 5…10 10…15 0.1…0.2 0.6…0.8 1.5…3 0.30…0.65
n = 400…750 rpm, four-stroke 8…15 4…9 8…13 0.05…0.15 0.5…0.7 1…1.1 0.45…0.7
n = 750…1500 rpm, two-stroke 10…40 3…7 4…8 0.05…0.1 0.5…0.7 1…1.2 0.5…0.75
n > 1500 rpm, two-stroke 15…30 3…6 3…7 0.04…0.08 0.6…0.8 1.2…2 0.5…0.9
As seen from the given system of equations, the first and

second stages of the combustion process are significantly
influenced by fuel vapourisation rate and mixture formation,
while the fuel post-injection stage is determined by the air
utilisation function in the engine cylinder ζa.c:
Φ 2
Φ –0.5( Φ )
z
ζa.c = 1–1.46(1– ζa.c0) Φz0z 2– e , (13) z0
√π
where Φz is the relative combustion duration; and ζa.c0 and
Φz0 are the coordinates of the minimum of the function
ζa.c = ζa.c(Φz0).
Table 1 provides typical values of the coefficients in the
fuel combustion model equations for different types of
engines, while Fig. 5 illustrates the influence of various tuning
parameters on the characteristics of heat release.
Fig. 7. Calculated diagrams of fuel evaporation rate. Medium-speed
diesel engine, 40% of MCR.
Considering variable fuel droplets
size during injection
As previously mentioned, Razleitsev used the average value

of the fuel droplet diameter d32 to predict fuel evaporation
and considered it to be constant during injection. In his latest
studies, Kuleshov considered variable fuel droplet size for the
ignition delay period [24], [25]. Obviously, this means that the
fuel evaporation constant bev, as well as the correction factor Y,
are also constant in the evaporation process. To examine the
applicability of simplified equations, consider Fig. 6-8, which
show the calculated fuel evaporation diagrams for two cases:
assuming a variable fuel droplets size across the injection
process and using the constant value of d32 averaged for the
injection period (d32 = const). As shown, assuming the variable Fig. 8. Calculated diagrams of fuel evaporation rate. High-speed
fuel droplets diameter makes a significant impact on fuel the diesel engine, triple injection, part load.
evaporation diagrams, which causes a further influence on Fuel combustion processes contribute to the emission of
fuel combustion and in-cylinder pressure build-up. toxic compounds in the exhaust gases of marine diesel engines.
These emissions primarily consist of CO, soot (and other
particulate matter), SOx, and NOx. Accurate prediction of
the buildup of toxic emissions is crucial for the application
of engine Digital Twins.
Considering toxic emission formation

IN the combustion processes
In order to predict the gas composition in different zones,

a two-zone combustion model is employed. This model
separates the fresh charge zone from the burned gases zone.
The composition of the burned gases is determined using
Professor Zvonov’s method, which assumes an 18-component
mixture consisting of O, O2, O3, H, H2, OH, H2O, C, CO, CO2,
Fig. 6. Calculated diagrams of fuel evaporation rate. Medium-speed CH4, N, N2, NO, NO2, NH3, HNO3, and HCN.
diesel engine, MCR.

The calculation of NOx concentration in the exhaust gas is The components of this equation are:
based on the Zeldovich mechanism for ‘thermal’ nitric oxide – t he kinetic soot formation rate (in the flame)
qfuel dx
(NO). This mechanism involves a series of three equations. (d[C]
dτ )kin = B1S V dt ;
The first equation is the most important, in terms of total – the core polymerisation of fuel droplets rate during fuel
NO formation kinetics. The equation for NO kinetics can be injection:
expressed as:
d[NO]
dτ
= K1p[N2][O]– K1r[NO][N]– K2r[NO][O]+
K3p[N][OH]– K3r[NO][H],
where the square brackets ‘[ ]’ express the volumetric and after fuel injection ends:
concentration of the corresponding matter, and K1p, K1r, K2p,
K2r, K3p, and K3r are constants for direct and reverse chemical
reactions.
The NO formation kinetic equation incorporates the first
and second equations, based on Zvonov’s approach:
(14) – the soot particles burning rate:

–
It should be noted that K4[O2][N2] = [NO]eq is the equilibrium (d[C]
dτ )burn = B3SkO2 √n · p · [C] ;
concentration of NO. – and the change in soot concentration rate due to cylinder
The conversion of the equation into volumetric fraction volume change:
units gives: (d[C] 6n dV
dτ )vol = B4S V dϕ .
In the above equations B1s, B’2s, B’’2s, B3s, and B4s are empiric
coefficients; δd is the droplet core size; Kev is the evaporation
constant; τinj.start and τinj.end are moments of time for injection
start and injection end; xinj.end is the burned fuel fraction for
(15) the moment of injection end; ndisp is the distibution constant
where p is the pressure in the cylinder (bar); R = 8.3144 J/ to consider fuel injection uniformity; k02is the oxidation
(mole·K) is the gas constant; and Tburned is the temperature of coefficient; and [C] is the volumetric soot concentration.
the burned gases.
The Arrhenius Law equations are used to calculate reaction
rate constants:
K = ATB exp (– RT
Ea
), (16)
where A and B are empirical coefficients; and Ea is the
activation energy.
For high-speed engines, the final concentration of CO in
exhaust gases is estimated as an equivalent concentration at
the combustion finish point. For medium-speed and low-
speed diesel engines, the following kinetic equation is used:
d[CO]
dτ
= K1C[CO] [OH] , (17)
where K1C = 7.1 ∙ 1012∙ e-32200RT is the reaction constant; and [CO]
and [OH] are the corresponding CO and OH concentrations.
Calculation of the formation of particulate matter in diesel
engines is performed using Razlejtsev’s approach. The equation
for the instantaneous volumetric soot concentration rate is
as follows:
d[C]
dτ
(
= d[C]
dτ
)kin+(d[C] ) –(d[C]
dτ pol
) –(d[C]
dτ burn
) .
dτ vol
(18) Fig. 9. Soot concentration and nitrogen oxide development in the cylinder.
Medium-speed engine, MCR.

The oxidation coefficient is considered using following It is essential to consider that the mere existence of such
equation: dependencies under actual operating conditions is insufficient,
as these relationships are derived for the proper state of the
(
kO2 = 1.8 1–(1–(1– α) ML02 ) x), (19) fuel system: the high-pressure fuel pump, high-pressure pipes,
and fuel injector.
where L0 is the stoichiometric amount of air (in kmole per Thus, it is necessary to be able to assess the characteristics
1 kg of fuel); and M2 is the amount of combustion products of fuel injection during the engine’s operation. To achieve this,
(in kmole per 1 kg of fuel). use of the following data is proposed:
The influence of the variable diameter of fuel droplets, • In the case of a research engine, sensors for fuel pressure
considering during the development of toxic emmissions, is are installed after the high-pressure fuel pump and directly
illustrated in Fig. 9. in front of the injector, as shown in Fig. 2. These sensors
allow for obtaining pressure diagrams at the corresponding
points, as well as a vibroacoustic sensor.
Discussion • For engines operating within a marine power plant, if the
installation of pressure sensors is not provided for by the
Therefore, it can be concluded that, the variable size of design, a vibroacoustic sensor is exclusively used, which
fuel droplets during injection is significant in the provided enables the acquisition of the moments of the beginning
combustion model, as it not only substantially affects and end of fuel injection [17].
the processes of vapourisation, mixture formation, and The data obtained from the sensors allow for obtaining the
combustion, but also the formation of toxic emissions, necessary characteristics of fuel injection and, importantly,
primarily nitrogen oxides and particulate matter. The taking into account their changes due to wear or malfunctions
proposed improvements in the mathematical model allow of the fuel apparatus. A detailed review of the algorithms
for a more accurate calibration with experimental data and, and results of similar data processing requires a separate
importantly for its Digital Twin application, do not lead to publication and represents a prospect for further research.
an increase in calculation time. Specifically indicated fuel consumption biNitrogen oxide
As an example, consider the calibration of a Digital Twin content at exhaust [NOx]Particulate matter content at exhaust
for a high-speed diesel engine with a cylinder diameter/stroke [Smoke]
of 120/120 mm, using the results of its comprehensive testing
under 27 operating modes, including the determination of
emissions of key pollutants (nitrogen oxides) and smoke Conclusions
opacity of the exhaust gases.
Table 2 illustrates the values of relative errors in determining Digital Twin technology represents a promising direction
the specific indicated fuel consumption bii, concentrations for enhancing the operational efficiency of marine internal
of nitrogen oxides [NOx], and smoke opacity [Smoke]. combustion engines. The advanced systems for monitoring the
Consideration of variable fuel droplet size during injection technical conditions and diagnosing engine malfunctions can
helped to obtain better cumulative accuracy, calculated with be developed, complying with the principles and requirements
the equation: of the fourth industrial revolution for modern transportation
Σδ2i systems. The performance-type Digital Twin, as the core
δΣ = N , (20) element of these systems, uses the data from engine sensors
to predict the virtual operating cycle, which gives additional
where N = 27 (the total number of measurements). information about engine parameters and helps to estimate
Table 2 shows that consideration of fuel droplet size the efficiency of the engine and detect possible malfunctions.
variation helped to simultaneously increase the accuracy of Technically, the monitoring system includes the server with
the prediction of bi – from δΣ = 5.7% to δΣ = 5.1%, [NOx] – the database for the engine operating parameters, transmitted
from δΣ = 6.2% to δΣ = 3.2% and [Smoke] from δΣ = 41.6% from the set of sensors. The data include slow-speed ‘static’ and
to δΣ = 17.8%. high-speed ‘dynamic’ parameters. The Digital Twin uses the
However, it is necessary to acknowledge that the proposed database in autonomous mode and is also capable of providing
refinement of the combustion model, which is thoroughly the malfunctions alarm or control inputs.
grounded from a theoretical standpoint, presents certain When developing a Digital Twin for an internal combustion
pragmatic challenges. It becomes evident that, in order to engine, it is important to reliably and accurately calculate
optimally harness the enhanced capabilities of the model, the development of the fuel combustion processes, as they
the characteristics of the fuel injection (in the form of multi- significantly impact both the engine’s efficiency and the level
parameter functions denoted as dσ/dφ = f(rpm, bmep)) of toxic emissions in exhaust gases. Considering the variation
are required, at the very least. The procurement of such in fuel droplet diameter during the injection process, allows for
data necessitates the execution of intricate and dedicated a substantial improvement in the accuracy of engine modelling
experimental investigations, typically carried out on specialised using the Digital Twin, as is shown for the set of 27 operational
test-benches. points of the high-speed diesel engine.

Tab. 2. Relative accuracy of Digital Twin predicted diesel engine parameters
Specifically indicated fuel consumption bi

d32 = const δΣ=5.7% d32 = var δΣ=5.1%
Ignition
Crank speed advance
40% load 70% load 100% load 40% load 70% load 100% load
10 14.1% 2.8% –0.7% 11.8% 1.2% –1.4%
1000 13 14.9% 8.6% 5.9% 13.3% 7.3% 6.0%
18 7.7% 4.5% 3.6% 6.8% 3.9% 3.2%
10 3.5% 1.2% –0.1% 1.8% –0.4% –1.5%
1400 13 9.5% 6.6% 2.0% 7.9% 5.3% 1.1%
18 2.7% 2.7% 3.6% 2.0% 2.1% 3.4%
10 –2.8% –3.3% 0.9% –4.5% –4.5% –1.7%
2200 13 1.6% –1.7% –2.6% 0.4% –2.4% –3.0%
18 –4.2% –2.6% –0.8% 4.7% –2.8% –2.3%
Nitrogen oxide content at exhaust [NOx]

Ignition
Crank speed d32 = const δΣ=6.2% d32 = var δΣ=3.2%
advance
10 –3.6% –1.6% –1.9% 1.7% 1.5% 1.6%
1000 13 –2.6% 0.4% 0.5% –0.8% –1.3% –0.8%
18 2.3% 1.4% 10.9% –0.8% –2.5% 3.2%
10 –6.6% 3.7% 2.1% –6.1% –1.3% 1.3%
1400 13 –2.2% 0.4% –1.0% 1.3% –0.2% –1.0%
18 –0.9% 0.8% –6.1% 0.9% –3.6% –2.9%
10 11.1% 21.2% –7.6% –10.0% 4.5% –6.4%
2200 13 –3.1% 1.8% –3.3% –2.0% –1.3% –1.0%
18 10.6% –3.0% –0.2% 3.6% –0.4% –0.3%
Particulate matter content at exhaust [Smoke]

Ignition
Crank speed d32 = const δΣ=41.6% d32 = var δΣ=17.8%
advance
10 –12.6% –1.9% –6.9% –23.8% –1.3% –4.2%
1000 13 13.9% 0.0% –2.3% 0.0% –0.1% 0.9%
18 33.7% 14.2% 41.8% 12.6% 15.7% 43.7%
10 –12.4% 7.5% 3.4% –24.4% 5.0% 7.9%
1400 13 –8.6% 23.4% –15.2% –21.2% 25.1% –10.6%
18 10.2% 17.2% 11.9% 0.1% 18.6% 15.6%
10 –24.5% 0.7% –4.4% –33.7% –7.9% –4.7%
2200 13 –20.6% –2.9% 3.8% –31.6% –11.0% 1.1%
18 –16.1% –11.6% 3.7% –24.3% –15.6% 1.1%
However, for the successful utilisation of the additional References

capabilities inherent to the fuel combustion model, it is
necessary to have the ability to conduct an evaluation of fuel 1. M. Grieves and J. Vickers, “Digital twin: Mitigating
injection characteristics during engine operation. This is unpredictable, undesirable emergent behavior in complex
proposed to be accomplished through the use of vibroacoustic systems,” Transdisciplinary perspectives on complex
sensors and, where feasible, pressure sensors within the fuel systems: New findings and approaches. 2017, pp. 85-113,
system, and represents the direction of future research. https://www.researchgate.net/profile/Michael-Grieves/

publication/306223791_Digital_Twin_Mitigating_ Transport,” in Systems, Decision and Control in Energy V.
Unpredictable_Undesirable_Emergent_Behavior_in_ Studies in Systems, Decision and Control, 2023, vol 481.
Complex_Systems/links/5aa54e1ea6fdccd544bc386f/ Springer, Cham., doi: 10.1007/978-3-031-35088-7_36.
Digital-Twin-Mitigating-Unpredictable-Undesirable-
Emerge. 12. S. Hautala, M. Mikulski, E. Söderäng, X. Storm and S. Niemi,
“Toward a digital twin of a mid-speed marine engine: From
2. S. Evans, C. Savian, A. Burns and C. Cooper, “Digital detailed 1D engine model to real-time implementation on
Twins for the Built Environment: An Introduction to the a target platform,” International Journal of Engine Research.
Opportunities,” Built Environmental News. 2019, https:// 2022, doi: 10.1177/14680874221106168.
www.theiet.org/media/8762/digital-twins-for-the-built-
environment.pdf. 13. S. Stoumpos, G. Theotokatos, C. Mavrelos and E. Boulougouris,
“Towards Marine Dual Fuel Engines Digital Twins — Integrated
3. D. Botín-Sanabria, A.-S. Mihaita, R. Peimbert-García, Modelling of Thermodynamic Processes and Control System
M. Ramírez-Moreno, R. Ramírez-Mendoza and J. Lozoya- Functions,” J. Mar. Sci. Eng. 2020, vol. 8, no. 3(200), doi: 10.3390/
Santos, “Digital Twin Technology Challenges and Applications: jmse8030200.
A Comprehensive Review,” Remote Sensing. 2022, vol. 14(6),
no. 1335, doi: 10.3390/rs14061335. 14. I. Asimakopoulos, L. Avendaño-Valencia, M. Lützen and N.
Rytter, “Data-driven condition monitoring of two-stroke marine
4. M. Singh, E. Fuenmayor, E. Hinchy, Y. Qiao, N. Murray and diesel engine piston rings with machine learning,” Ships and
D. Devine, “Digital Twin: Origin to Future,” Appl. Syst. Innov. Offshore Structures. 2023, doi: 10.1080/17445302.2023.2237302.
2021, vol. 4, no. 36, doi: 10.3390/asi4020036.
15. O. Bondarenko and T. Fukuda, “Development of a diesel engine’s
5. L. Li, S. Aslam, A. Wileman and S. Perinpanayagam, “Digital digital twin for predicting propulsion system dynamics,” Energy.
Twin in Aerospace Industry: A Gentle Introduction,” 2020, vol. 196, doi:10.1016/j.energy.2020.117126.
IEEE Access. 2022, vol. 10, pp. 9543-9562, doi: 10.1109/
ACCESS.2021.3136458. 16. R. Varbanets, O. Fomin, V. Píštěk, V. Klymenko, D. Minchev,
A. Khrulev, V. Zalozh and P. Kučera, “Acoustic method
6. M. Xia, H. Shao, D. Williams, S. Lu, L. Shu and C.W. de Silva, for estimation of marine low-speed engine turbocharger
“Intelligent fault diagnosis of machinery using digital twin- parameters,” Journal of Marine Science and Engineering.
assisted deep transfer learning,” Reliability Engineering & 2021, vol. 3, no. 9, doi: 10.3390/jmse9030321.
System Safety. 2021, vol. 215, doi:10.1016/j.ress.2021.107938.
17. R. Varbanets, O. Shumylo, A. Marchenko, D. Minchev,
7. S. Choi, J. Woo, J. Kim and J. Lee, “Digital Twin-Based V. Kyrnats, V. Zalozh, N. Aleksandrovska, R. Brusnyk and
Integrated Monitoring System: Korean Application Cases,” K. Volovyk, “Concept of vibroacoustic diagnostics of the
Sensors. 2022, vol. 22, no. 5450, doi: 10.3390/s22145450. fuel injection and electronic cylinder lubrication systems
of marine diesel engines,” Polish Maritime Research. 2022,
8. D. Zhong, Z. Xia, Y. Zhu and J. Duan, “Overview of predictive vol. 29, no. 4, pp. 88-96, doi: 10.2478/pomr-2022-0046.
maintenance based on digital twin technology,” Heliyon. 2023,
vol. 9, no. 4, doi: 10.1016/j.heliyon.2023.e14534. 18. S. Neumann, R. Varbanets, D. Minchev, V. Malchevsky and V.
Zalozh, “Vibrodiagnostics of marine diesel engines in IMES
9. A.T. Hoang, A.M. Foley, S. Nižetić, Z. Huang, H.C. Ong, GmbH systems,” Ships and Offshore Structures. 2022, doi:
A.I. Ölçer, V.V. Pham and X.P. Nguyen, “Energy-related 10.1080/17445302.2022.2128558.
approach for reduction of CO2 emissions: A critical strategy
on the port-to-ship pathway,” Journal of Cleaner Production. 19. O. Yeryganov and R. Varbanets, “Features of the fastest pressure
2022, vol. 355, doi:10.1016/j.jclepro.2022.131772. growth point during compression stroke,” Diagnostyka. 2018,
vol. 19, no. 2, pp. 71-76, doi: 10.29354/diag/89729.
10. O. Melnyk, O. Sagaydak, O. Shumylo and O. Lohinov,
“Modern Aspects of Ship Ballast Water Management and 20. D. Minchev, R. Varbanets, N. Alexandrovskaya and
Measures to Enhance the Ecological Safety of Shipping,” L. Pisintsaly, “Marine diesel engines operating cycle simulation
in Systems, Decision and Control in Energy V. Studies in for diagnostics issues,” Acta Polytechnica. 2021, vol. 61, no. 3,
Systems, Decision and Control, Springer ed. 2023, vol. 481, pp. 428-440, doi: 10.14311/ap.2021.61.0435.
Cham, doi: 10.1007/978-3-031-35088-7_39.
21. D. Minchev, O. Gogorenko, R. Varbanets, Y. Moshentsev, V.
11. O. Onishchenko, A. Bukaros, O. Melnyk, V. Yarovenko, Píštěk, P. Kučera, O. Shumylo and V. Kyrnats, “Prediction of
A. Voloshyn and O. Lohinov, “Ship Refrigeration System centrifugal compressor instabilities for internal combustion
Operating Cycle Efficiency Assessment and Identification engines operating cycle simulation,” Proceedings of the
of Ways to Reduce Energy Consumption of Maritime Institution of Mechanical Engineers, Part D: Journal of

Automobile Engineering. 2023, vol. 237, no. 2-3, pp. 572-
584, doi: 10.1177/09544070221075419.
22. Н. Ф. Разлейцев, Моделирование и оптимизация

процесса сгорания в дизелях, Харьков: Вища школа,
1980, p. 169.
23. А. Ф. Шеховцов, Ф. И. Абрамчук and В. И. и. д. Крутов,

Процессы в перспективных дизелях, Харьков: Основа,
1992, p. 352.
24. L. Grekhov, K. Mahkamov and A. Kuleshov, “Optimization

of Mixture Formation and Combustion in Two-Stroke OP
Engine Using Innovative Diesel Spray Combustion Model
and Fuel System Simulation Software,” SAE. 2015, 2015-01-
1859, doi: 10.4271/2015-01-1859.
25. A. Kuleshov, K. Mahkamov, A. Kozlov and Y. Fadeev,

“Simulation of dual-fuel diesel combustion with multi-zone
fuel spray combustion model,” Proceedings of the ASME
2014 Internal Combustion Engine Division Fall Technical
Conference. 2014, pp. 1-13, doi: 10.1115/ICEF2014-5700.

10.2478/pomr-2023-0062
AN EXPERIMENTAL STUDY OF THE EFFECTS OF CYLINDER

LUBRICATING OILS ON THE VIBRATION CHARACTERISTICS
OF A TWO-STROKE LOW-SPEED MARINE DIESEL ENGINE
Gang Wu 1
*
Guodong Jiang 1
Changsheng Chen 2
Guohe Jiang 1
Xigang Pu 1
Biwen Chen 2
1
Merchant Marine College, Shanghai Maritime University, Shanghai, China
2
Shanghai Marine Equipment Research Institute, Shanghai, China
* Corresponding author: wugang@shmtu.edu.cn (Gang Wu)
Abstract
Two-stroke, low-speed diesel engines are widely used in large ships due to their good performance and fuel economy.
However, there have been few studies of the effects of lubricating oils on the vibration of two-stroke, low-speed diesel
engines. In this work, the effects of three different lubricating oils on the vibration characteristics of a low-speed engine are
investigated, using the frequency domain, time-frequency domain, fast Fourier transform (FFT) and short-time Fourier
transform (STFT) methods. The results show that non-invasive condition monitoring of the wear to a cylinder liner in
a low-speed marine engine can be successfully achieved based on vibration signals. Both the FFT and STFT methods
are capable of capturing information about combustion in the cylinder online in real time, and the STFT method also
provides the ability to visualise the results with more comprehensive information. From the online condition monitoring
of vibration signals, cylinder lubricants with medium viscosity and medium alkali content are found to have the best
wear protection properties. This result is consistent with those of an elemental analysis of cylinder lubrication properties
and an analysis of the data measured from a piston lifted from the cylinder after 300 h of engine operation.
Keywords: two-stroke; low-speed marine diesel engine; cylinder lubricating oils; vibration characteristic; condition monitoring
INTRODUCTION can have direct effects on the hull, machinery, crew, and
passengers when it occurs in ships. The main source of engine
The internal combustion (IC) engine is a classical rotating vibration is an imbalance in the force and moment created by
power machine that has been widely employed for transportation the inertial force of its moving parts, although another source
and power production [1]. In particular, the two-stroke low- is the side thrust and overturning moment generated by the
speed diesel engine is a unique type of IC engine that has been in-cylinder pressure of fuel combustion. Further factors include
used in many medium and large ships due to its advantages the influence of lubricating oils [3, 4], fuels [5-8], and engine
such as multi-fuel feasibility, fuel economy, large output power, faults [9, 10], which can all affect engine vibration. The engine
and durability [2]. With the increasing emphasis on ship safety, contains many internal components, such as the piston-cylinder
more and more researchers are focusing on the hazards to assembly, which require a good lubricating oil to run; hence,
reliability posed by engine vibration. Harmful engine vibration the characteristics of the lubricating oil, and its quality and

performance, will have a significant impact on the behaviour wear, whereas if too much alkali is present, inorganic calcium
of the engine, especially for two-stroke low-speed marine diesel carbonate will form hard deposits that cannot be burned out,
engines with complex structures and large dimensions. leading to abrasive wear. With the trend towards low-sulphur
In general, lubricating oil is an oil-like liquid that is applied fuel, there is a lack of experience in terms how to address the
between moving surfaces and can carry out the functions of issues of lubricity and cleanliness between the piston and
reducing friction and wear. It also has the ability to eliminate cylinder liner. In addition, the traditional method of inspecting
shock loads, and can ensure cooling, sealing, etc. Thus, the piston and cylinder liner relies on manual disassembly of
lubricating oil requires several properties, such as a suitable the piston-cylinder assembly, which is not efficient.
viscosity, flash point, and alkali value, which are related to the In recent years, vibration and noise signals have been
functions mentioned above. Prior research [11] has confirmed shown to be useful for assessing the condition of engines,
the importance of the properties of the lubricant, as it affects such as combustion processes, valve faults, and fuel injection
not only the engine itself but also attached components such behaviours, and can be used for fault detection in diesel
as turbochargers, meaning that it is very important in terms of engines. Researchers have already begun to investigate
the transient response of the engine. The viscosity-temperature engine performance and optimal fuel selection based on
characteristics of lubricating oil reveal the relationship between engine vibration signals. Ahmad Taghizadeh-Alisaraei [12]
viscosity and temperature, where the lower the temperature, the designed a systematic experiment to evaluate the minimum
higher the viscosity. This relationship is strongly related to the engine vibration for seven engine speeds and nine biodiesel
performance of the engine, as a high viscosity for the lubricating fuel blend ratios by calculating the root mean square (RMS)
oil will increase the friction loss of the engine, especially in and comparing the amplitudes of the vibration signals. Ahmet
a low-temperature environment. It is therefore important to Çalık [14] measured the vibration characteristics of an engine
develop suitable grades of engine lubricating oil with different when hydrogen was added to biodiesel. The RMS was used to
viscosities. The American Society of Automotive Engineers compare the vibrations of different fuel blends. Their results
(SAE) grading system is widely used internationally to represent showed that biodiesel fuel could reduce engine vibration,
the viscosity of lubricating oil. and hydrogen could reduce the vibration further. Similarly,
Lubricating oil with different properties can produce different ethanol-diesel blended fuel has also been investigated for
levels of friction and wear in an engine, which can lead to a diesel engine [15], and the RMS and kurtosis were used to
different vibration characteristics and different amounts of wear scrutinise the performance of the engine.
to metals from the lubricating oil used. In research conducted by In other studies, airborne acoustic signals have been used to
Ahmad Taghizadeh-Alisaraei, it was found that the replacement assess the lubrication quality of the engine. Albarbar et al. [16]
of lubricating oil helps to reduce engine vibration, as it can studied the effects of lubricating oil based on structure-borne
reduce the friction between components such as bearings, acoustic signals, and found that the amplitude of the spectral
piston cylinders, etc. [12]. One potential reason for this is that components was proportional to the engine speed and load.
degradation of the lubricating oil and the presence of wear The RMS values of the signals were found to be influenced by
debris in the unreplaced lubricating oil lead to greater engine the lubricant conditions.
vibration. Recent studies have revealed further information on Some newly developed signal analysis and processing
the effects of fuel type on the status of lubrication oil. Nantha methods have also been used as effective tools for detecting
Gopal et al. [13] studied the effects of biodiesel fuel on the piston scratching faults. In Ref. [15], the short-time Fourier
surface wear to cylinder liners, and quantified the cylinder transform (STFT) was developed to characterise various sources
wear of the engine by comparing the surface roughness of the of engine vibration. Through an STFT analysis, the authors
cylinder liner as well as the concentration of wear debris in successfully identified engine knocking cycles when the ethanol
the lubricant samples from the engine. The results revealed concentration in diesel fuel exceeded 8%. Moosavian et al. [17]
significant degradation of the lubricating oil in a biodiesel investigated the vibration signature of a diesel engine under
blended fuel engine. A further analysis of the oil revealed healthy and fault operating conditions, using the fast Fourier
the presence of wear metals in the used oil, which were also transform (FFT) and continuous wavelet transform (CWT)
considered to be related to degradation. Hence, the problem methods to explore the approximate frequency band caused by
of degradation is critical for lubrication oil, as it can cause a piston scratching fault. The results showed that the scratching
a significant reduction in the kinematic viscosity and the flash fault excited the 3–4.7 kHz frequency band [17]. Omar et al.
point of the lubricating oil. [18] carried out an experimental study in which FFT and STFT
For a two-stroke low-speed marine diesel engine, the type were used to compare the vibrations of a liquefied petroleum
of lubrication required for the cylinder liner is unique. A larger gas (LPG) dual-fuel engine and a base diesel engine. The results
cylinder bore and a longer cylinder stroke require a greater showed that the vibration of the LPG dual-fuel engine was lower
injection of lubricating oil to the cylinder. In addition, matching than for the diesel engine.
the sulphur level of heavy fuel oil (HFO) to the cylinder Although the studies described above show that the use
lubricating oil is a critical issue. This requires that the cylinder of vibration signal is promising in terms of monitoring the
lubricating oil has an appropriate alkali value: if too little alkali tribological behaviours of engine cylinders, there are few
is present, the acids in the combustion products will not be studies of the monitoring of two-stroke low-speed marine
neutralised and will attack the piston liner, causing corrosive diesel engines fuelled by HFO, which are vital to prevent the

gradual degradation of cylinder lubrication oil and to avoid zinc compound additive, etc., but they had different SAE grades
any unnecessary wear and corrosion of the cylinder liner. and total base numbers (TBNs). The SAE grade represents the
In this paper, we improve on the findings described above viscosity of the oil. The detailed specifications of these lubricating
by using a more powerful analysis based on the RMS, FFT, oils are shown in Table 2. It is worth noting that although the
and STFT methods to characterise the engine vibration signals viscosity of recipe B is as same as that of recipe C (both are
acquired for cylinder lubricating oils with different SAE grades SAE 40), they have different TBNs.
and base numbers, thereby establishing a more accurate and
Tab. 2. Specifications of the lubricating oils used for testing
reliable indicator for monitoring the tribological behaviours
of the piston ring and cylinder liner of a two-stroke low- Lubricating oil SAE grade TBN
speed marine diesel engine. To further verify the diagnostic Recipe A SAE100 TBN 70
performance of the vibration signal responses extracted using
optimised FFT and STFT spectra, three formulas for cylinder Recipe B SAE 40 TBN 40
lubricating oils were sampled for elemental analysis after 300 h Recipe C SAE 40 TBN 25
of steady-state engine operation, and the carbon deposits on
the surface of the pistons were compared before and after these Fig. 1 presents a schematic diagram of the experimental
300 h of operation. The main objectives were to explore the setup used for vibration testing, which shows the connections
effects of different cylinder lubricating oils on the wear to the between the accelerometer, proximity sensor, power supply, data
engine cylinder, and to explore the vibration characteristics of acquisition system, load control system, and a laptop computer
a two-stroke low-speed marine diesel engine fuelled by HFO. with a measuring system. A dynamometer was employed to apply
the load to the engine via a load control system. The accelerometer
was mounted on the engine head, and a proximity sensor was
MATERIALS AND METHODS coupled to the crankshaft to measure its angle. The signals from
both transducers were transferred to a data acquisition system
(INV3062-C2(L)), and the output data cable was connected to
EXPERIMENTAL SETUP AND DATA ACQUISITION the data acquisition system and a laptop computer to transfer
the data. Finally, the data were analysed by a measuring system
The experiments were performed on a two-stroke low-speed in the laptop computer. The electricity for the data acquisition
marine diesel engine, manufactured by MAN B&W corporation, system was provided by a power supply.
which was installed in the laboratory of Shanghai Maritime
University (SMU), China. This type of low-speed diesel engine
is widely used in large ocean-going vessels, and is a typical
example of such an engine. The engine was fuelled with low-
sulphur-content heavy fuel oil (0.5% S) during the experiments.
The main technical specifications of this low-speed engine are
shown in Table 1.
Tab. 1. Technical specifications of the engine used in the experiments
Item/parameter Details
Brand MAN B&W
Model 6S35ME-B9
6-cylinder, water-cooled, direct
Engine type injection, crosshead type
Combustion order 1-5-3-4-2-6
Fig. 1. Schematic diagram of the experimental setup.
Bore & stroke 350 × 1550 mm
The vibration signals were measured with a tri-axial vibration
Compression ratio 17:1
accelerometer (made by Coinv, China). Detailed specifications
Displacement 894 L of the transducers are shown in Table 3. As can be seen from
Rated output power 3750 kW @ 142 rpm Fig. 1, the accelerometer was mounted on the engine cylinder
cover, and the vibrations in three directions (x, y, z) were all
Maximum indicated pressure 180 bar
acquired by the accelerometer simultaneously. The vibration
Valve opening pressure 380 bar signals were measured under the same engine load condition
for each of the three cylinder lubricating oils. The angle of the
Three different formulas of cylinder lubricating oil (recipes A, engine crankshaft was recorded by the proximity sensor during
B, and C) were prepared and used. These three cylinder oils all the vibration measurements. The coordinate system used for
contained the same percentages of additives to the base oil, which both the accelerometer and the engine is shown to the lower
included compound ester, ester base, amide, phosphate ester, right of Fig. 1.

Fig. 2. Installation layout of the experimental equipment.
Tab. 3. Accelerometer specifications
each axis for all experiments was calculated using Eq. (1) [9]:
Item/parameter Details N
1
Type INV9832-50
xRMS = N Σ xi2 (1)
i=1
Sensitivity 100 mV/g where xRMS (m/s ) is the value of the RMS for the calculated
2
Measuring range 0.4–12,000 Hz signal.

To compare the total vibration for each cylinder, the values
Weight 12 g
of RMStotal were calculated for three axes (x, y, and z) using
Size 19×19×19 mm3 Eq. (2) [15]:
Temperature −50 to 120 °C
RMSTotal = √RMSx2 + RMSy2 + RMSz2 (2)
The locations at which the vibration sensors were installed
are shown in Fig. 2. where RMSTotal (m/s2) is the total vibration acceleration, and
In order to approximate the engine operating conditions for RMSx, RMSy and RMSz are the RMS values for the accelerations
the navigation of a real ship, the experiments included three along the x, y, and z axes, respectively.
different engine loads (25%, 40%, and 50%). Each sampling was
performed after the engine speed had stabilised over 6 h. The Kurtosis of acceleration signal
data acquisition period was approximately 30 s, and the sampling The kurtosis is a parameter used to describe the sharpness
frequency was 5 kHz for each record; these values of the sampling of a signal distribution. In general, the sharper the signal
time and frequency allowed sufficient engine operating cycles centroid, the higher the kurtosis. The magnitude of its value
to be recorded. The other parameters of the engine, such as its can be regarded as the degree to which the signal is abnormal.
speed, output power, and torque, are shown in Table 4. To analyse the kurtosis, its value was calculated using Eq. (3) [9]:
Tab. 4. Additional engine parameters ΣNi=1(xi –x–)4
xKurtosis = (N–1)σ 4 (3)
Engine load Engine speed Output power Torque
25% 98 rpm 812.5 kW 91.64 kN·m To compare the total vibration in each cylinder based on
the kurtosis, Kurtosistotal was derived from the values for the
40% 105 rpm 1300 kW 116.13 kN·m
x, y and z axes using Eq. (4) [15]:
50% 113 rpm 1625 kW 136.40 kN·m
KurtosisTotal = √Kurtx2 + Kurty2 + Kurtz2 (4)
ANALYSIS METHODS
where KurtosisTotal is the total vibration acceleration, and Kurtx,
In order to compare the differences between the vibration Kurty and Kurtz are the values of the kurtosis for accelerations
signals in the time domain, the values of the RMS, kurtosis, along the x, y, and z axes, respectively.
mean, skewness, standard deviation, and squared deviation of
the signals were considered. Mean of acceleration signals
For the frequency and time-frequency domain analyses, the The mean is the sum of all the data in a set divided by the
FFT and STFT were used. These analyses were all performed total number of data points. It is used to reflect the general
using MATLAB (2019 b) and Origin software. overall level of data, and can be calculated using Eq. (5) [9]:
N
RMS of acceleration signal x– = N1 Σ xi (5)
i=1
The RMS represents the strength of the signal, and is also an
expression of the average energy of the signal. The RMS along

Skewness of acceleration signals the time-domain features become, and the frequency-domain
The skewed direction and degree of the signals can be features become more insignificant; conversely, the longer the
described by the skewness parameter. The skewness value window, the clearer the frequency-domain features and the
determines whether the signal has a positive, negative, or zero more indistinct the time features.
skew (normal distribution), and is calculated by Eq. (6) [9]:
N –3
xSkewness = Σ(N–1)σ
i=1(xi –x)
3 (6) RESULTS AND DISCUSSION
Standard deviation of acceleration signals The objectives of this study were to compare the effects of
The standard deviation of the arithmetic square root of the three different lubricating oils on a two-stroke low-speed marine
squared deviation, which indicates the degree of dispersion diesel engine, and to explore the vibration of the engine under
between samples of vibration data, is defined by Eq. (7) [10]: low to medium load conditions. The results of the experimental
study are therefore presented in four sections relating to the
ΣNi=1(xi –x–)2
σ= N (7) time domain, the frequency domain, a time-frequency domain
analysis of the vibration signal, and an elemental spectra
Squared deviation of acceleration signals analysis.
The squared deviation of signals represents the degree
of fluctuation of the signal around the mean; it can also be
understood as the degree of vibration signal shock that exists.
It is calculated using Eq. (8):
Σ (xi – x–)2 (8)
N
δ = N1
i=1
Fast Fourier transform

The feature parameters mentioned above are quantitative
assessments of the time domain of signals, whereas an analysis
of the frequency domain of the signal is also required. The
FFT can be used to convert a signal from the time domain to
the frequency domain, so that the frequency-domain content
of the signal can be obtained. The FFT is defined as shown in Fig. 3. A complete engine cycle in the time domain
Eq. (9) [15]:
n
X(k) = Σ X(j)Wn(j–1)(k–1) (9) TIME DOMAIN ANALYSIS OF VIBRATION SIGNALS
j=1
where n is the length of signal X, and Wn = e(–2πi)/n. The lubricating oil not only provides lubrication and
reduces friction between the cylinder and piston ring, but
Short-time Fourier transform also participates in the combustion process. The vibration of
In the STFT, a signal is split into multiple segments using an IC engine mainly arises from the gas force generated by
a window function (in our analysis, the Hamming window the combustion in the cylinders, and a prior study [19] was
function was used). The FFT is then applied to calculate each conducted to explore the in-cylinder combustion process of
segment of the signal, and the spectrograms obtained from a low-speed engine using the CFD method. The usage of different
the FFT are arranged along the time axis, to obtain the time- lubricating oils is also a crucial factor affecting the vibration in
frequency diagram. Hence, one of the advantages of the STFT the engine. To investigate the detailed information for the time
over the FFT is that the time and frequency information can domain, we consider the z 3 s segment of the signal in Fig. 3,
both be retained simultaneously, as the time information is lost where a complete engine cycle is marked by red lines. This signal
when the FFT is used. Another advantage is that the STFT can represents the conditions with lubricating oil recipe B, at an
be applied to non-stationary signals, which is not possible for engine speed of 98 rpm under a load of 25%, along the z-axis.
the FFT. The formula for the STFT is given in Eq. (10) [17]: As shown in the figure, the firing order of the cylinders is 1-5-
+∞ 3-4-2-6. The acceleration of the vibration signal in cylinder 4
STIF(τ, f ) = ∫ x(t)h(t – τ)e–j2πft (10)
–∞
has the maximum value, while that of cylinder 1 shows the
minimum. The reason for this is that the accelerometer was
where h is a window function. f and τ denote the frequency mounted on the head of cylinder 4, whereas cylinder 1 was
and time variables, respectively. farthest from the accelerometer. The acceleration values for
However, the STFT also has certain limitations, as the size the other cylinders decrease with increasing distance from
of the window cannot change with frequency. Thus, the time cylinder 4. The recording period for the vibration signal was
and frequency resolution cannot be optimised at the same 30.1056 s, containing 150,528 samples. Hence, for an engine
time in the STFT. The shorter the window, the more obvious speed of 98 rpm, it can be calculated that the duration of

a complete cycle under the 25% load condition is approximately a 40% load. To enable a comparison of the total vibration of
0.6122 s, containing 3,061 samples. Similarly, from the speeds the diesel engine, the RMSTotal of the vibration signal was
in Table 4, we can calculate the time for a complete cycle under calculated for the x, y, and z vibrations using Eq. (2). There are
the 40% load condition as 0.5714 s, and that under the 50% clear similarities in Fig. 4(b) between the data for the RMS and
load condition as 0.5310 s. RMSTotal for recipe B under the 25% and 40% load conditions.
Before carrying out a frequency domain signal analysis, we However, under a load of 50%, an increase in RMSTotal is observed
calculated the time domain feature parameters (RMS, RMSTotal, for recipes A, B and C compared to the RMS. This suggests that
kurtosis, KurtosisTotal, mean, skewness, standard deviation, the vibrations along the other two axes are increased under the
and squared deviation) for the vibration signals for the three 50% load condition due to the centrifugal inertia force.
recipes of lubricating oil for 15 complete engine cycles at loads Figs. 4(c) and 4(d) show a high degree of similarity between
of 25%, 40%, and 50%. Figs. 4 and 5 display scattergrams of the data point distributions of the kurtosis and KurtosisTotal.
the calculated results, showing 15 data points for each recipe Under a 25% load, the data points are scattered, whereas when
under each load condition. the load is increased, the data points become more concentrated.
Firstly, the value of the RMS for each axis was calculated This indicates that the vibration signal for this marine diesel
using Eq. (1) and the calculation results for the z-axis are shown engine is comparatively sharp under low load conditions, but
in Fig. 4a. It can be clearly seen that the RMS values for recipe shows improvement as the load increases. We can also see that
B are low for all three load conditions, and the data points for the data points for recipe B are concentrated at lower values
this recipe are relatively concentrated. This indicates that the compared to the other two recipes of lubricating oil. The results
marine diesel engine experiences low vibration along the z-axis for the kurtosis showed a trend that was consistent with the
under low to medium load when lubricated with lubricating oil results for the RMS. Thus, recipe B for the lubricating oil gives
based on recipe B. It also suggests that the medium-level TBN better vibration performance for this diesel engine.
of cylinder oil is better matched to its low-sulphur property and
gives better engine performance. Recipe A also gave low RMS
values under the 25% and 40% load conditions, in a similar
way to recipe B. However, under the 50% load condition, there
was an increase in some of the RMS values compared to the
previous low load conditions. For recipe C, the RMS values
show fluctuations under each of the three loads, meaning that
TBN values that are too high or too low are not conducive to
stable engine vibration.
Fig. 5. Plots of feature parameter values for three engine loads for three
recipes of lubricating oil: (a) mean; (b) skewness; (c) standard deviation;
(d) squared deviation.
Fig. 5 shows the values of the mean, skewness, standard
deviation, and squared deviation. For each of the three loads, the
mean values for recipe A are almost all near zero, while the mean
values for recipe B remain near zero only for the 40% and 50%
load conditions. It is not hard to observe from Fig. 5(a) that the
mean values of all three lubricating oils are relatively steady and
Fig. 4. Plots of feature parameter values for three engine loads for three recipes remain near zero under the 40% load condition. Fig. 5(b) shows
of lubricating oil: (a) RMS; (b) ; (c) kurtosis; (d) . that under the 40% and 50% load conditions, recipe B has low
The RMS values for each recipe of lubricating oil are low values for the RMS, kurtosis, and mean, whereas its skewness
under the 40% load condition. The reason for this is related to values are almost all negative. In contrast, the data points for
the characteristics of this diesel engine, as different lubricating recipe A are uniformly distributed on both sides of zero, and
oils have little effect on the intensity of engine vibrations under do not vary greatly between the load conditions. Furthermore,

it is clear that the data points are most concentrated under the beings around an engine [18]. It can be seen from the results
40% load condition for all three recipes of lubricating oil. In in Figs. 6–8 that the low-frequency vibration range of this
Figs. 5(c) and 5(d), the standard deviation data are fairly close engine shows a decreasing trend as the load increases, which
to those of the RMS, and both have almost identical scatter is beneficial to the life of the engine and the comfort of the
distributions. Meanwhile, the scattergram distribution of the crew and passengers on board.
squared deviation is also in accordance with both of the above, The range 0.75–1.3 kHz contains the dominant frequencies
despite the different values. of the engine, which are induced by the shock generated by
the combustion pressure. Our results show that there are no
FREQUENCY AND TIME-FREQUENCY DOMAIN significant changes in the maximum value in this subdomain
ANALYSES OF VIBRATION SIGNALS for recipes A and C. For example, the maximum values of
recipe C in this range are close to 0.330 m/s2. However, the
In order to further analyse the effects of each oil recipe on maximum values of recipe B show a certain degree of reduction,
the vibration of this two-stroke low-speed diesel engine, the with a decline of 27.7%. This finding indicates that although
vibration signals along the z-axis under three load conditions lowering the alkalinity to a moderate level facilitates lightening
were transformed using FFT. Ten consecutive working cycles the intensity of the engine combustion, a TBN level that is too
for each treatment condition were considered for the FFT low does not significantly improve the vibration from engine
analysis. combustion.
Fig. 6. FFT spectra for recipe A under three typical engine loads: Fig. 7. FFT spectra for recipe B under three typical engine loads:
(a) 25%, (b) 40%, (c) 50%. (a) 25%, (b) 40%, (c) 50%.
Figs. 6–8 show FFT spectra for vibration signals under In the range 1.3–1.75 kHz, all three recipes for the lubricating
three different engine loads for the three recipes of lubricating oil show different trends with increasing load. However,
oil. It can be observed that the vibration signatures of the under the 50% load condition, their maximum values are all
diesel engine change when different cylinder lubricating at a relatively low level of below 0.34 m/s2. The changes in the
oils are used, from Fig. 6 to Fig. 8. The changed frequency maximum values for recipes A and B are similar in the high-
domain of 0–2.5 kHz can be divided into four subdomains: frequency subdomain (1.9–2.5 kHz), as both increase with
0.12–0.38 kHz, 0.75–1.3 kHz, 1.3–1.75 kHz, and 1.9–2.5 kHz. rising loads. The increase in the maximum value for recipe A is
In the range 0.12–0.38 kHz, which is the low-frequency region 198%, and for recipe B it is 161%. It is notable that the maximum
of engine vibration, the accelerations all decrease to varying value under 25% load is 0.426 m/s2 for recipe C, which is higher
extents for all three lubricating oils as the load increases, and than for its principal frequency domain. However, under the
especially their maximum values. Due to its high viscosity 40% load condition, the maximum value is only 0.1 m/s2,
and alkaline nature, the decrease in the maximum value for while the value is 0.352 m/s2 under the 50% load condition.
recipe A at a higher load is greater than that for recipes B and C, We conclude that under low to medium loads, using lubricating
with a decline of 60.2%. The maximum values for recipes B oil with recipe C causes high-frequency vibration fluctuations
and C are fairly similar, as they have the same SAE 40 grade. in the engine. This also provides some evidence that a TBN
It is known that the low-frequency components of vibration level that is too low may affect the detergency between the
are the main contributors to noise and discomfort for human piston and liner.

lower peaks can be seen, indicating the vibration of the
remaining cylinders; the amplitude for the cylinder on which the
tri-axis vibration sensor was mounted on is the largest, and the
amplitudes for the other cylinders decrease towards both sides
of the engine. The STFT method can be used to analyse engine
vibration caused by combustion and friction. This approach
is an excellent visualisation tool for monitoring the condition
of diesel engines, and can easily provide information about
engine vibration and combustion. Since combustion occurs
inside the cylinder, it cannot be observed readily. Hence, the
state of combustion which is non-visualized can be observed
using the STFT method.
As can be seen from Fig. 9a, at 40% engine load, the
maximum vibration amplitude of the low-speed marine engine
is around 1.5 kHz for all three oils, where the vibration is mainly
caused by combustion. Recipe A shows three consecutive
periods of unusually high amplitude in the frequency band
1.3–1.75 kHz; there are also two continuous engine cycles
where the amplitude is extremely low. In Fig. 9(c), the peak of
each cycle in the 1.9–2.5 kHz range for recipe C is comparable
to the values for recipes A and B, but the vibration of the other
Fig. 8. FFT spectra for recipe C under three typical engine loads: cylinders is much lower than for recipes A and B. What can
(a) 25%, (b) 40%, (c) 50%.
be clearly seen in the 10 cycles for recipe B is better stability
Furthermore, from the time-domain parameters described in all four subdomains. This further corroborates the findings
above, it can be seen that the vibration data for this engine with for the time domain, and indicates that recipe B has the best
these three lubricating oils are relatively centralised under a 40% vibration characteristics of the three lubricating oils under
load, which implies good stability of the engine. In order to 40% load conditions.
further explore the effects of the three oil recipes on the engine
vibration characteristics, the STFT method was applied to analyse LINER WEAR ANALYSIS
only the vibration signals under the 40% load condition. Unlike
the FFT method, the STFT method involves time information, The carbon deposits on the surface of each piston were
meaning that the differences between various engine cycles can compared before and after 300 h of steady-state engine operation.
be observed in the STFT spectra. The signals analysed for each These deposits covered the surface of each piston with a similar
of the above segments all had a length of 5.714 s, and contained thickness, and were mainly composed of the combustion
10 consecutive engine cycles. products from the HFO and cylinder lubricating oils as well
In Fig. 9, each STFT diagram contains 10 obvious spectrum as piston-cylinder frictional wear particles, of which the major
peaks, representing the 10 engine cycles. In each cycle, several elements were C, O, S, Fe, and Al.
Fig. 9. STFT diagrams for lubricating oils under 40% load: (a) recipe A; (b) recipe B; (c) recipe C.

In order to study the main components of the frictional wear effects of the lubricating oils on engine vibration, in order
particles, the elements in the lubricating oil were measured to select the optimum lubricating oil to reduce wear, and to
through an oil analysis. Several typical elements were selected investigate the vibration characteristics of the low-speed engine
to compare their contents for evaluation. The cylinder of the under typical load conditions. A triaxial vibration transducer
marine diesel engine was mainly made of iron-carbon alloy, was placed at planned locations. The vibration signals obtained
meaning that Fe was the main metallic element. Al and Si were in the experiments were analysed using various time domain
the main components of the piston, while Mn, Cu, and Ni were characteristics, FFT and STFT. The metallic elements in the
also contained in the piston material. Since the three lubricating cylinder lubricating oils were also measured via an oil analysis
oils have the same additive ratio but different formulas, they before and after operation. The conclusions from the experiment
varied in terms of the content of each metallic material before can be summarised as follows.
service. We therefore calculated the difference value (D-value) The vibration characteristics of the two-stroke, low-speed
for each major element before and after service. From Fig. 10, marine diesel engine operating on heavy fuel oil were first
it can be seen that the D-values of Fe are 75.2×10-6 for recipe investigated, and it was found that the characteristic time-
B, 80×10-6 for recipe A, and 88.2×10-6 for recipe C, meaning domain parameters of the vibration signal were relatively stable
that the D-value for recipe B is the lowest. For the other five at low engine loads, due to the reduced friction frequency. The
elements, the D-values are also the lowest for recipe B, followed alkalinity and viscosity characteristics of the cylinder lubricant
by recipe A and then recipe C. This indicates that recipe B has had the least effect on the engine vibration under a 40% load.
the best anti-wear performance of the three types of oil. The influence of the cylinder lubricant on engine vibration
Recipe C has the same SAE grade as recipe B but a lower gradually increased as the load increased.
TBN value, meaning that its alkali level is too low to completely Under typical ship propulsion loads, recipe B was found to
neutralise the acids of combustion; the un-neutralised acid will have the most concentrated and stable data distributions for
attack the piston liners, resulting in corrosive wear and more RMS, kurtosis, mean, skewness, standard deviation, and squared
wear debris during the operation of the pistons. Thus, a cylinder deviation of the three lubricating oils. This indicates that the
lubricating oil with medium alkali values provides better anti- medium level alkaline value and medium level viscosity of the
wear properties than one with a low alkali value. Recipe A cylinder oil are more suitable for low-sulphur heavy oil and
has a somewhat higher D-value than recipe B. One reason for cause less wear of cylinder liners and less vibration of the engine.
this is that its excess alkali is in the form of inorganic calcium Real-time FFT and STFT methods were successfully
carbonate that cannot be burned, causing hard deposits and used for non-invasive monitoring of the cylinder liner wear
three-body abrasive wear between the piston and cylinder. The of a marine engine. The sub-ranges of the FFT graphs also
high viscosity of recipe A will also increase the friction loss of show that moderately alkaline cylinder oils improve engine
the engine, and may increase the wear of the piston-cylinder smoothness, while excessively low alkaline levels can worsen
assembly to a certain extent. In brief, a cylinder lubricating oil vibration in the engine combustion chamber. The STFT method
with a medium level of viscosity and a medium alkali value has was used as a new way to visually analyse the cylinder under its
the best anti-wear properties. operating conditions in terms of time and frequency, and the
results agreed well with those of a time domain analysis and
offline elemental analysis.
This study has presented the results of a vibration data
analysis of a two-stroke low-speed marine diesel engine in
healthy condition under low to medium operating conditions,
which can serve as a data reference for future fault monitoring.
Fault diagnosis of a two-stroke low-speed marine diesel engine
will be explored in our future work.
Author Contributions
Gang Wu: Conceptualization, Methodology, Investigation,

Writing - original draft & editing, Funding acquisition. Guodong
Jiang: Investigation, Experimentation, Data curation, Writing -
Fig. 10. Difference values for each major element before and after service. original draft & editing. Changsheng Chen: Conceptualization,
Methodology. Guohe Jiang: Writing - review. Xigang Pu:
Experimentation. Biwen Chen: Writing - review.
CONCLUSION
Acknowledgement
In this experimental study, the vibration signals for
three different cylinder lubricating oils were measured The authors disclose receipt of the following financial support
using a six-cylinder low-speed marine diesel engine (MAN for the research, authorship, or publication of this article: This
B&W 6S35ME-B9). The main objectives were to analyse the work was supported by the Science & Technology Commission

of Shanghai Municipality and Shanghai Engineering Research 9. A. Moosavian, G. Najafi, B. Ghobadian, M. Mirsalim, S. M. Jafari,
Center of Ship Intelligent Maintenance and Energy Efficiency and P. Sharghi, „Piston scuffing fault and its identification in
under Grant 20DZ2252300. an IC engine by vibration analysis,” Appl. Acoust., vol. 102,
pp. 40-48, 2016, https://doi.org/10.1016/j.apacoust.2015.09.002.
Conflicts of Interest
10. S. M. Ramteke, H. Chelladurai, and M. Amarnath, „Diagnosis
The author(s) declare no potential conflicts of interest with and classification of diesel engine components faults using time–
respect to the research, authorship, and/or publication of this frequency and machine learning approach,” J. Vib. Eng. Technol.,
article. vol. 10, no. 1, pp. 175-192, 2021, https://doi.org/10.1007/
s42417-021-00370-2.
REFERENCES 11. E. G. Giakoumis, „Lubricating oil effects on the transient

performance of a turbocharged diesel engine,” Energy,
1. A. M. K. P. Taylor, „Science review of internal combustion vol. 35, no. 2, pp. 864-873, 2010, https://doi.org/10.1016/j.
engines,” Energ. Policy, vol. 36, no. 12, pp. 4657-4667, 2008, energy.2009.08.009.
https://doi.org/10.1016/j.enpol.2008.09.001.
12. A. Taghizadeh-Alisaraei, B. Ghobadian, T. Tavakoli-Hashjin,
2. M. Yang et al., „Matching method of electric turbo and S. S. Mohtasebi, „Vibration analysis of a diesel engine
compound for two-stroke low-speed marine diesel engine,” using biodiesel and petrodiesel fuel blends,” Fuel, vol. 102,
Appl. Therm. Eng., vol. 158, 2019, https://doi.org/10.1016/j. pp. 414-422, 2012, https://doi.org/10.1016/j.fuel.2012.06.109.
applthermaleng.2019.113752.
13. K. Nantha Gopal and R. Thundil Karuppa Raj, „Effect of
3. J. Kałużny et al., „Reducing friction and engine vibrations pongamia oil methyl ester–diesel blend on lubricating oil
with trace amounts of carbon nanotubes in the lubricating degradation of di compression ignition engine,” Fuel, vol. 165,
oil,” Tribol. Int., vol. 151, 2020, https://doi.org/10.1016/j. pp. 105-114, 2016, https://doi.org/10.1016/j.fuel.2015.10.031.
triboint.2020.106484.
14. A. Çalık, „Determination of vibration characteristics of
4. W. Li, Y. Guo, X. Lu, X. Ma, T. He, and D. Zou, „Tribological a compression ignition engine operated by hydrogen enriched
effect of piston ring pack on the crankshaft torsional vibration diesel and biodiesel fuels,” Fuel, vol. 230, pp. 355-358, 2018,
of diesel engine,” Int. J. Engine Res., vol. 16, no. 7, pp. 908-921, https://doi.org/10.1016/j.fuel.2018.05.053.
2014, https://doi.org/10.1177/1468087414556134.
15. A. Taghizadeh-Alisaraei and A. Rezaei-Asl, „The effect of
5. Ö. Böyükdipi, G. Tüccar, and H. S. Soyhan, „Experimental added ethanol to diesel fuel on performance, vibration,
investigation and artificial neural networks (ANNs) based combustion and knocking of a CI engine,” Fuel, vol. 185,
prediction of engine vibration of a diesel engine fueled with pp. 718-733, 2016, https://doi.org/10.1016/j.fuel.2016.08.041.
sunflower biodiesel – NH3 mixtures,” Fuel, vol. 304, 2021,
https://doi.org/10.1016/j.fuel.2021.121462. 16. A. Albarbar, F. Gu, A. Ball, and A. Starr, „Internal combustion
engine lubricating oil condition monitoring based on vibro-
6. K. Çelebi, E. Uludamar, and M. Özcanlı, „Evaluation of fuel acoustic measurements,” Insight, vol. 49, no. 12, pp. 715-718,
consumption and vibration characteristic of a compression 2007, https://doi.org/10.1784/insi.2007.49.12.715.
ignition engine fuelled with high viscosity biodiesel and hydrogen
addition,” Int. J. Hydrogen Energ., vol. 42, no. 36, pp. 23379- 17. A. Moosavian, G. Najafi, B. Ghobadian, and M. Mirsalim,
23388, 2017, https://doi.org/10.1016/j.ijhydene.2017.02.066. „The effect of piston scratching fault on the vibration behavior
of an IC engine,” Appl. Acoust., vol. 126, pp. 91-100, 2017,
7. K. Çelebi, E. Uludamar, E. Tosun, Ş. Yıldızhan, K. Aydın, and M. https://doi.org/10.1016/j.apacoust.2017.05.017.
Özcanlı, „Experimental and artificial neural network approach
of noise and vibration characteristic of an unmodified diesel 18. F. K. Omar, M. Y. E. Selim, and S. A. Emam, „Time and
engine fuelled with conventional diesel, and biodiesel blends frequency analyses of dual-fuel engine block vibration,”
with natural gas addition,” Fuel, vol. 197, pp. 159-173, 2017, Fuel, vol. 203, pp. 884-893, 2017, https://doi.org/10.1016/j.
https://doi.org/10.1016/j.fuel.2017.01.113. fuel.2017.05.034.
8. G. Chiatti, O. Chiavola, and F. Palmieri, „Vibration and 19. X. Zhou, T. Li, N. Wang, X. Wang, R. Chen, and S. Li, „Pilot
acoustic characteristics of a city-car engine fueled with diesel-ignited ammonia dual fuel low-speed marine engines:
biodiesel blends,” Appl. Energ., vol. 185, pp. 664-670, 2017, A comparative analysis of ammonia premixed and high-
https://doi.org/10.1016/j.apenergy.2016.10.119. pressure spray combustion modes with CFD simulation,”
Renew. Sust. Energ. Rev., vol. 173, 2023, https://doi.org/10.1016/j.
rser.2022.113108.

10.2478/pomr-2023-0063
EXPLORATION OF A MODEL THERMOACOUSTIC

TURBOGENERATOR WITH A BIDIRECTIONAL TURBINE
Volodymyr Korobko 1
Serhiy Serbin 1
*
Huu Cuong Le 2
1
Admiral Makarov National University of Shipbuilding, Mykolaiv, Ukraine
2
Institute of Maritime, Ho Chi Minh City University of Transport, Viet Nam
* Corresponding author: serbin1958@gmail.com (Serhiy Serbin)
Abstract
The utilisation of the thermal emissions of modern ship power plants requires the development and implementation of
essentially new methods of using low-temperature waste heat. Thermoacoustic technologies are able to effectively use low-
temperature and cryogenic heat resources with a potential difference of 500–111 K. Thermoacoustic heat machines (TAHMs)
are characterised by high reliability, simplicity and environmental safety. The wide implementation of thermoacoustic
energy-saving systems is hampered by the low specific power and the difficulties of directly producing mechanical work.
An efficient approach to converting acoustic energy into mechanical work entails the utilisation of axial pulse bidirectional
turbines within thermoacoustic heat engines. These thermoacoustic turbogenerators represent comprehensive systems that
consist of thermoacoustic primary movers with an electric generator actuated by an axial-pulse bidirectional turbine. The
development of such a thermoacoustic turbogenerator requires several fundamental issues to be solved. For this purpose,
a suitable experimental setup and a 3D computational fluid dynamics (CFD) model of a thermoacoustic engine (TAE)
with bidirectional turbines were created. The research program involved conducting physical experiments and the CFD
modelling of processes in a TAE resonator with an installed bidirectional turbine. The boundary and initial conditions for
CFD calculations were based on empirical data. The adequacy of the developed numerical model was substantiated by the
results of physical experiments. The CFD results showed that the most significant energy losses in bidirectional turbines
are manifested in the output grid of the turbine.
Keywords: waste heat recovery; ship power plant; thermoacoustics; thermoacoustic engine; bidirectional turbine
INTRODUCTION The International Maritime Organization (IMO), in

accordance with the provisions of the United Nations Sustainable
Maritime transport is an important component of the global Development Goal (SDG) 13 [1]–[3], introduced requirements
economy, as it provides the bulk of freight traffic. Marine heat that direct the general trend in shipping towards a significant
engines consume carbon-based recoverable fuel, producing limitation of the volume and composition of emissions from
significant amounts of heat emissions of various potentials ship power plants, including diesel and gas turbine units [4]–[8].
that harm the environment. These emissions include fuel The task of the decarbonisation of marine energy has
combustion products, such as CO2, SOn, NOx, volatile organic become vital; it aims to minimise the consumption of carbon
compounds (VOСs), carbon particles, etc. fuels, which is a key factor in reducing greenhouse gas (GHG)

emissions. The result of these measures was the introduction of rotor. Therefore, it can be considered that such thermoacoustic
efficient dual-fuel engines that use new types of fuel, including turbine generators (TATGs) are the most rational solution for
cryogenic fuels. These measures led to a significant decrease in low-temperature ship WHRSs.
the temperature of thermal emissions (Table 1). Obviously, the use of bidirectional turbines in thermoacoustic
Under such conditions, traditional ship thermal emission devices is an appropriate solution. However, it should be noted
utilisation schemes, based on the classic water Rankine cycle that OWC units differ significantly from TAE units in terms of
(WRC), become ineffective. It is these circumstances that lead the design and parameters of the working environment, and
to the use of other energy-saving technologies, such as the therefore the direct use of existing approaches in the design of
organic Rankine cycle (ORC) system [9] ]–[11] and thermo- such turbines is problematic. The existing problem is a complex
chemical conversion [12]. task and requires a set of additional studies.
A promising method for the utilisation of the low-temperature
thermal emissions of ship power plants (SPPs) may be the use
of thermoacoustic technologies [13], [14]. FUNCTIONING CONDITIONS
There are two types of thermoacoustic machines: OF BIDIRECTIONAL TURBINES
thermoacoustic engines (TAEs) and thermoacoustic refrigerators
(TARs), or heat pumps. Thermoacoustic engines (movers) It is known that OWC plants are open systems that operate at
implement a direct process – converting thermal energy from atmospheric pressure (Pm = Pamb). In these systems, the working
external sources into an acoustic form. A TAR performs the medium, air, undergoes oscillatory motion. In the flowing part
opposite process – it consumes acoustic energy and converts it of the OWC, the amplitude of these oscillations’ motion can
into the heat with different potential. reach several metres, while the frequency of oscillations does
not exceed 0.2 to 1 Hz:
Tab. 1. Temperatures of waste heat carriers of modern SPPs
us
2S ICE 4C ICE PEM FC ζ = ρa = (0.05–0.1)Pm , (1)
Heat carrier/source 2πfρ √χRT
Temperature, K
where ρ is the density of the medium in kg/m3; χ is the adiabatic
Waste gases of ICE 490–530 500–690 –
index; T is the temperature in K; and R is the universal gas
Air charger 400–490 380–470 – constant.
(LT PEM) 333–363 TAEs are closed systems in which noble gases are the working
Cooling system liquid 355–360 360–370 (HT PEM) 450–493 fluid, the internal pressure Pm can reach 0.3–3.0 MPa and the
LNG fuel 111 111 111 frequency of the acoustic wave is f = 50–150 Hz.
NH3 fuel 240 240 240
Accordingly, in thermoacoustic devices, the amplitude of
the oscillatory motion of the acoustic wave ζTAE is much smaller
In ship power systems, TAEs can utilise high-potential heat than ζOWC(Fig. 1).
sources with temperatures ranging from 330–550 K or higher,
such as emissions from internal combustion engines (ICEs) or
electrochemical generators like solid oxide fuel cells (SOFCs)
and polymer electrolyte membrane fuel cells (PEMFCs). On
the other hand, low-potential thermal sources might be the
ambient environment or cryogenic fluids such as liquefied
hydrogen (LH2), natural gas (LNG) or ammonia (NH3), with
temperatures from 4–240 K [15]. The widespread application
of thermoacoustic systems in practice is hindered by the
complexity of the efficient conversion of acoustic energy into
mechanical work, a low specific power density and the lack of
practical experience. Issues related to the development of low-
temperature thermoacoustic systems were considered in [16].
In terms of renewable energy, we are familiar with wave Fig. 1. Comparison of the amplitudes of the oscillatory motion
power plants that utilise oscillating water columns (OWCs), of the medium in an OWC and a TAE
which incorporate bidirectional turbines like the Wells turbine It is obvious that the hydrodynamic processes in the flow
and impulse bidirectional turbines [17]–[19]. The application of channels of OWC systems are significantly different from the
bidirectional turbines in thermoacoustic heat recovery systems processes in TAE resonators. In a thermoacoustic apparatus
has also been addressed in previous studies [19]–[23]. under high-frequency oscillations, the influence of inertial
Bidirectional turbines are the most efficient transducers for factors on the hydrodynamic parameters of the medium should
ship thermoacoustic waste heat recovery systems (WHRSs) increase significantly. In addition, there is a significant difference
when it comes to generating mechanical work and driving between the designs of the flow parts of these bidirectional
electric generators. They are capable of transforming the energy turbines. In the case of an OWC, large-volume collectors are
of acoustic oscillation into the rotational motion of the turbine used. In thermoacoustic devices, the resonators have much

Fig. 2. Experimental turbine generator assembly: 1 – turbine rotor, 2 – stator, 3 – generator, 4 – fairing, 5 and 6 – turbine axes, 7 and 8 – fasteners
smaller dimensions. In this regard, it is possible to expect the electric grid. The characteristics of bidirectional turbines in
formation of stable oscillating structures in the TATG resonator, a unidirectional flow were studied using the experimental
which will introduce additional excitations and inhomogeneities installation, the scheme of which is shown in Fig. 3.
at the input to the rectifier of the turbine. They will create a novel Tab. 2. Design parameters of bidirectional turbines
energy-consuming mechanism that should be confined on both Element Value
sides of the bidirectional turbine. In view of these circumstances, Internal diameter of TAE resonator, mm 46
it can be considered appropriate to conduct a complex of studies
Outer diameter of turbine stator, mm 76
to obtain an in-depth understanding of these issues.
Outer blade diameter of turbine, mm 75
Purpose of research. Considering the disparities in the Turbine rotor diameter, mm 50
working processes between OWCs and TAEs, it is sensible to Blade width of rotor, mm 20
carry out research focused on understanding the peculiarities of Number of rotor blades 24
hydrodynamic processes within the TATG resonator to identify Width of stator blades, mm 16
the most rational design solutions. Research should involve
Number of stator blades 18
both physical experiments and numerical calculations using
computational fluid dynamics (CFD) procedures, which requires Differential thermostabilised pressure sensors of MPXV NXP
the creation of appropriate equipment and methods of research. semiconductors were used to measure the pressure drop along
the turbine and on the entrance lemniscate. The equipment of
the experimental stands and the available microprocessor-based
Experimental setup data acquisition system (DAS) provided multiple measurements
of sensor readings in each of the modes, which reduced the
The investigation was conducted utilising an experimental random error.
bidirectional turbine. This turbine was designed and The STM32F407VGT6 microcontroller, which has a 32-bit
manufactured in accordance with the guidelines outlined in architecture and an operating frequency of 168 MHz, was used
[24]. A 3D model of an experimental bidirectional turbine to measure and register signals from pressure sensors, current
was developed in the CAD software developed by SolidWorks. sensors, voltage sensors, etc.
The turbine’s components, including the rotor, guide grids and The equipment of the test stands ensured multiple
fairings, were produced using 3D printing techniques and plastic measurements of sensor readings in each of the modes, which
material, as depicted in Fig. 2. Detailed information about the reduced the random error.
dimensions of the elements of the experimental bidirectional
turbine is provided in Table 2.
The rotor of the turbine was rigidly coupled to a three-phase
brushless electric generator of the Sankyo F2JGL type. The
output voltage from the generator was fed to a diode converter
made of Schottky diodes, and a P517-M laboratory rheostat or
a set of precision resistors was used as a load.
The rotation speed of the turbine was monitored by an SDS
1074CFL digital oscilloscope, which was connected to one of
the windings of the generator.
Fig. 3. Diagram of the experimental stand for the study of the hydraulic resistance
Such a scheme made it possible to simultaneously control of the bidirectional turbine: 1 – lemniscate; 2 – inclined differential
the speed of rotation of the turbine rotor, the root-mean-square pressure gauge; 3 – turbine; 4 – fan; 5 – LATR; 6 – DAS oscilloscope
(RMS) voltage of the generator VRMS and the current in the SDS 1074CFL; 7 – diode bridge; 8 – ammeter; 9 – voltmeter; 10 – resistors

The polling frequency of the pressure sensor controller by
the STM32F407VGT6 controller was 10 kHz, and the bit rate
was 32 bits.
A digital 4-channel oscilloscope, SDS 1074CFL, was used for
the visual control of the operation of the TAE and the progression
of the experiments, as a duplicating the measurement and
registration of research information. Communication with the
DAS and the oscilloscope was ensured using the USB interface.
This microprocessor-based control and measurement system
is discussed in more detail in [25].
An important task of the experimental research was to assess
the operability of a model sample of a pulsed bidirectional
turbine and its suitability for further experiments. All failings
that were identified at the preparation stage were fixed. Fig. 5. Turbine rotor revolutions depending on the speed of the oncoming flow
Taking into account instrument errors and errors of It can be seen that the turbine rotation frequency also
approximation of expressions, the total measurement error depends on the generator load. The maximum speed of rotation
does not exceed 2.5%. The error of determining the power of under the conditions of the experiment reached 5300 rpm. As
the generator, when using the existing DAS, does not exceed 2%. can be expected, the maximum frequency of rotation of the
turbine occurred when there was no load, in the idle regime.
Additionally, during experimental studies, data were obtained
EXPERIMENT RESULTS that made it possible to determine its power depending on the
speed of rotation. The results of these calculations are presented
At this stage of the experiments, the characteristics of the in Fig. 6.
turbogenerator exclusively in the longitudinal flow at different
air flows rates were studied. The results of the experiments, that
is, the pressure drops of the bidirectional turbine, are shown
in Fig. 4.
Fig. 6. The electrical power generated by the experimental turbine

generator at various rotor speeds and loads
The minimum load mode of the generator was simulated

using a resistor rated at 500 ohms. The results of the tests showed
Fig. 4. Dependence of hydraulic resistance of bidirectional turbines the suitability of the turbogenerator model for further research.
on flow velocity under different loads
The load on the turbogenerator was changed using a set of
precision resistors with different resistance values Ri. Data on CFD SIMULATION RESULTS
the hydraulic resistance of a turbine with a fixed rotor were
also obtained. For a detailed study of hydrodynamic processes in the TAE
Research results have shown that a turbine with a fixed rotor resonator, a series of computational experiments was carried
has minimal flow resistance, which is quite natural. For a turbine out. These studies were carried out with the help of various
with a free rotor, as the load decreases, the flow resistance software products, which made it possible to compare the results
increases. Furthermore, the data of these studies were used to obtained and assess the adequacy of the modelling.
verify the correctness of the results of the CFD modelling. The In the ANSYS package [26], the channel geometry was
results of measuring the frequency of rotation of the turbine created using the ANSYS Design Modeler and Blade Modeler
rotor depending on the speed of the oncoming flow are shown programs, and the mesh was created using the ANSYS Turbo
in Fig. 5. The main task of these experiments was to test the Grid program (Fig. 7); the analysis of the results was carried out
performance of the turbine in long-term operating modes and using the CFD-Post software module. The features of various
to assess the stability of external indicators. mathematical models that can be used in the CFD approach
are described in detail in [27]–[31].

Fig. 7. Experimental sample of a bidirectional turbine and its CFD model: a) ANSYS Design Modeler, b) SolidWorks 3D model
In the first stage of research, in order to check the correctness

of the construction of the CFD calculation models and create
a time-efficient calculation scheme, test calculations were carried
out. Considering the complexity of the task, the calculations were
restricted to the fixed turbine rotor. In these experiments, the
conditions of full-scale experiments to measure the resistance of
a bidirectional turbine in a unidirectional flow were simulated.
The boundary conditions were set in accordance with the
parameters of the physical experiment, and the standard k-e
turbulence model was used. Fig. 8 shows the results of the CFD
simulation of the hydraulic resistance of a resonator section
with a bidirectional turbine.
It can be seen that the results of the CFD calculations and
physical experiments coincide satisfactorily, since the difference
between them does not exceed 10%. The existing deviations Fig. 8. Hydraulic resistance of IDT specimen: comparison of experimental
data with CFD simulation results
in the results can be explained by the difficulty of determining
and taking into account the roughness of the surface of the The vortex structures in the flow behind the bidirectional
bidirectional turbines and by some design deviations that were turbines are more effectively visualised through the results
due to the capabilities of the existing 3D printer. presented in Figs. 12 and 13.
In further calculations, the hydrodynamic structure of the These figures demonstrate that within the resonator of the
flow in the resonator with bidirectional turbines was studied. thermoacoustic generator, swirl flows developed, wherein
The calculations were made for the speed range of 3–25 m/s, the tangential velocity component equals nearly 60% of the
which fully covers the range of amplitude values of the oscillating longitudinal component. These persistent circulating vortex
velocity that could be achieved during the experiments. The structures possess substantial energy accumulation capabilities
complexity of the hydrodynamic picture is shown in Figs. 9–11. and are rather stable.
Fig. 9. Distribution of static pressure in the resonator channel with bidirectional turbines at different flow rates

Fig. 10. Distribution of static pressure in the resonator channel with bidirectional turbines at different flow rates
Fig. 11. Velocity distribution across the resonator cross-section
secondary vortex structures of the Hertler-Taylor type, which

also represent an energy-consuming mechanism.
Fig. 12. Current lines in a resonator with an experimental turbine;

the rotor is stationary
It can be argued that these phenomena in conditions

of oscillating motion should lead to a non-uniform flow
distribution in front of the straightening blades of bidirectional
turbines, which is an undesirable factor. When designing such
bidirectional turbines, measures should be taken to reduce these
effects. In addition, the presence of tangential currents on the
Fig. 13. Distribution of circumferential velocity along the length of the resonator
concave surface of the resonator will lead to the formation of with bidirectional turbines: a) velocity of 5 m/s; b) velocity of 10 m/s

CONCLUSIONS 2. International Maritime Organization, “Initial IMO GHG
strategy.” [Online]. Available: https://www.imo.org/en/
To perform the research program, a suitable experimental MediaCentre/HotTopics/Pages/Reducing-greenhouse-gas-
setup was developed; it includes a test worktable and emissions-from-ships.aspx. [Accessed: Oct. 15, 2023].
a microprocessor data acquisition and recording system.
A prototype of a turbogenerator equipped with a pulsed 3. International Maritime Organization, “Note by the
bidirectional turbine was fabricated, and a numerical International Maritime Organization to the UNFCCC
counterpart was established for CFD simulations. Talanoa Dialogue.” [Online]. Available: https://unfccc.int/
First, experimental evaluations were conducted under sites/default/files/resource/250_ IMO%20submission_
straight-line medium movement conditions with varying Talanoa%20Dialogue_April%202018.pdf. [Accessed: Oct. 15,
generator loads. Hydraulic resistance measurements of the 2023].
turbine model within the resonator were sourced empirically.
The results demonstrated that peak pressure losses transpire 4. International Maritime Organization, “IMO. Low carbon
during the idling phase at the generator’s minimal load. Under shipping and air pollution control.” [Online]. Available: http://
such conditions, rotor rotational speeds reached up to 5500 rpm. www.imo.org/en/MediaCentre/HotTopics/GHG/Pages/
A CFD framework was constructed to represent the default.aspx. [Accessed: Oct. 15, 2023].
combination of a bidirectional turbine and a thermoacoustic
engine resonator. This model facilitated the exploration of 5. International Maritime Organization, “Fourth IMO GHG
turbine characteristics both for a stationary rotor and during study 2020 executive-summary.” [Online]. Available: https://
the rotor’s rotation. wwwcdn.imo.org/localresources/en/OurWork/Environment/
The hydraulic resistance computations for the stationary-rotor Documents/ Fourth%20IMO%20GHG%20Study%20
turbine aligned closely with empirical results, deviating by less 2020%20Executive-Summary.pdf. [Accessed: Oct. 15, 2023].
than 10%. This concurrence validates the utility of the developed
CFD model for subsequent investigations. Simulations, executed 6. Finnish Marine Industries, “Journey to a carbon-free world:
across various software packages, consistently yielded analogous Introducing the NYK SUPER ECO SHIP 2050.” [Online].
outcomes. Available: https://meriteollisuus.teknologiateollisuus.fi/en /
Numerical analyses revealed the emergence of pronounced ajankohtaista/news/journey-carbon-free-world-introducing-
tangential and radial secondary currents immediately behind nyk-super-eco-ship-2050. [Accessed: Oct. 15, 2023].
the bidirectional turbine within the resonator. These secondary
vortex patterns introduce augmented energy expenditures and 7. H. Shi, Q. Zhang, M. Liu, K. Yang, and J. Yuan, “Numerical
require minimisation. Given the derived results, optimising the Study of the Ejection Cooling Mechanism of Ventilation
design of the bidirectional turbine’s guide apparatus, as well as for a Marine Gas Turbine Enclosure,” Polish Maritime
refining the profiling of the shroud and the resonator, emerges Research, Vol. 29, No. 3, pp. 119–127, 2022, doi: org/10.2478/
as a reasonable solution. pomr-2022-0032.
In analysing the current results, it is very important to
consider that the simulations were focused on a small two- 8. T. Niksa-Rynkiewicz, A. Witkowska, J. Głuch, and
way turbine with an output power of only a few watts. However, M. Adamowicz, “Monitoring the Gas Turbine Start-Up
for shipboard thermoacoustic WHRSs, the turbine generators Phase on a Platform Using a Hierarchical Model Based
should provide power at levels of approximately 102 to 103 on Multi-Layer Perceptron Networks,” Polish Maritime
kW. For such significantly larger structures, the occurrence of Research, Vol. 29, No. 4, pp. 123–131, 2022, doi: 10.2478/
so-called large-scale effects is likely. In addition, the question of pomr-2022-0050.
the construction of TATG resonators and their overall aggregate
location is open. 9. E.-L. Tsougranis and D. Wu, “A feasibility study of organic
In future efforts, the priority should be the integration of Rankine cycle (ORC) power generation using thermal and
innovative design solutions aimed at mitigating the appearance cryogenic waste energy on board an LNG passenger vessel,”
of secondary radial flows in the resonators of high-power International Journal of Energy Research, Vol. 42, No. 9,
thermoacoustic generators. pp. 3121–3142, July 2018, doi: 10.1002/er.4047.
10. M. E. Mondejar, J. G. Andreasen, L. Pierobon, U. Larsen,

REFERENCES M. Thern, and F. Haglind, “A review of the use of organic
Rankine cycle power systems for maritime applications,”
1. N. Olmer, B. Comer, B. Roy, X. Mao, and D. Rutherford, Renewable and Sustainable Energy Reviews, Vol. 91, pp. 126–
“Greenhouse gas emissions from global shipping.” [Online]. 151, 2018, doi: 10.1016/j.rser.2018.03.074.
Available: https://theicct.org/wp-content/uploads/2021/06/
Global-shipping-GHG-emissions-2013-2015_ICCT- 11. T. Hoang, “Waste heat recovery from diesel engines based
Report_17102017_vF.pdf. [Accessed: Oct. 15, 2023]. on Organic Rankine Cycle,” Applied Energy, Vol. 231. 2018,
doi: 10.1016/j.apenergy.2018.09.022.

12. O. Cherednichenko, S. Serbin, and M. Dzida, “Application of Systems, Decision and Control, Vol. 171, pp. 469–485, 2019,
thermo-chemical technologies for conversion of associated doi: 10.1007/978-3-030-00253-4_20.
gas in diesel-gas turbine installations for oil and gas floating
units,” Polish Maritime Research, Vol. 26, No. 3, pp. 181–187, 24. T. Kloprogge, “Turbine design for thermo-acoustic generator,”
Sep. 2019, doi: 10.2478/pomr-2019-0059. Master’s thesis, Aeronautical Engineering, Hogeschool.
Holland Delft, 2012. Available: https://bioenergyforumfact.
13. G. W. Swift, “Thermoacoustic engines,” J. Acoust. Soc. Am., org/sites/default/files/2018-06/5.%20Turbine%20 Design
Vol. 84, No. 4, pp. 1145–1180, 1988. %20for%20a%20Thermo-acoustic%20Generator.pdf.
[Accessed: Oct. 15, 2023].
14. G. W. Swift, “Thermoacoustics: A unifying perspective for
some engines and refrigerators,” Acoust. Soc. Am., 2002. ISBN 25. Y. Kondratenko, O. Korobko, and V. Korobko, “Microprocessor
0-7354-0065-2. system for thermoacoustic plants efficiency analysis based
on a two-sensor method,” Sensors & Transducers, Vol. 24,
15. L. M. Qi, P. Lou, K. Wang, et al., “Characteristics of onset and Aug. 2013. Available: https://www.academia.edu/95466184/
damping in a standing-wave thermoacoustic engine driven Microprocessor_System_for_Thermoacoustic_Plants_
by liquid nitrogen,” Chin. Sci. Bull., Vol. 58, pp. 1325-1330, Efficiency_Analysis_Based_on_a_Two_Sensor_Method.
2013, doi: 10.1007/s11434-012-5214-z. [Accessed: Oct. 15, 2023].
16. Z. Yang, V. Korobko, M. Radchenko, and R. Radchenko, 26. ANSYS, Inc., ANSYS Fluent Theory Guide. ANSYS, Inc., 2013.
“Improving thermoacoustic low-temperature heat recovery
systems,” Sustainability (Switzerland), Vol. 14, No. 19, art. 27. S. I. Serbin, I. B. Matveev, and G. B. Mostipanenko, “Plasma-
No. 12306, 2022, doi: 10.3390/su141912306. assisted reforming of natural gas for GTL: Part II - Modeling
of the methane-oxygen reformer,” IEEE Trans. Plasma
17. T. K. Das, P. Halder, and A. Samad, “Optimal design of air Sci., Vol. 43, No. 12, pp. 3964–3968, 2015, doi: 10.1109/
turbines for oscillating water column wave energy systems: TPS.2015.2438174.
A review,” Int. J. Ocean Clim. Syst., Vol. 8, No. 1, pp. 37–49,
2017, doi: 10.1177/1759313117693639. 28. I. Matveev, S. Serbin, T. Butcher, and N. K. Tutu, “Flow structure
investigations in a Tornado combustor,” in 4th International
18. A. F. O. Falcao and J. C. C. Henriques, “Oscillating-water- Energy Conversion Engineering Conference, AIAA2006-4141,
column wave energy converters and air turbines: A review,” Vol. 2, 2006, pp. 1001–1013, doi: 10.2514/6.2006-4141.
Renewable Energy, 2015, doi: 10.1016/j.renene.2015.07.086.
29. O. Cherednichenko, S. Serbin, and M. Dzida, “Investigation
19. A. Thakker and F. Hourigan, “Modeling and scaling of the of the combustion processes in the gas turbine module of
impulse turbine for wave power applications,” Renewable an FPSO operating on associated gas conversion products,”
Energy, Vol. 29, No. 3, pp. 305–317, 2004, doi: 10.1016/ Polish Maritime Research, Vol. 26, No. 4, pp. 149–156, 2020,
S0960-1481(03)00253-2. doi: 10.2478/pomr-2019-0077.
20. D. Liu, Y. Chen, W. Dai, et al., “Acoustic characteristics of 30. S. Serbin, K. Burunsuz, M. Dzida, J. Kowalski, and D. Chen,
bi-directional turbines for thermoacoustic generators,” “Investigation of ecological parameters of a gas turbine
Front. Energy, Vol. 16, pp. 1027–1036, 2022, doi: 10.1007/ combustion chamber with steam injection for the floating
s11708-020-0702-3. production, storage, and offloading vessel,” International
Journal of Energy and Environmental Engineering, Vol. 13,
21. M. A. Elhawary, A. H. Ibrahim, A. S. Sabry, and E. Abdel- No. 3, pp. 873–888, 2022, doi: 10.1007/s40095-021-00433-w.
Rahman, “Experimental study of a small scale bi-directional
axial impulse turbine for acoustic-to-mechanical power 31. I. B. Matveev, N. V. Washchilenko, and S.I. Serbin, “Plasma-
conversion,” Renewable Energy, 2020, doi: 10.1016/j. Assisted Reforming of Natural Gas for GTL: Part III - Gas
renene.2020.05.162. Turbine Integrated GTL,” IEEE Trans. Plasma Sci., Vol. 43,
No. 12, pp. 3969–3973, 2015, doi: /10.1109/TPS.2015.2464236.
22. C. Iniesta, J. L. Olazagoitia, J. Vinolas, and J. Aranceta,
“Review of travelling-wave thermoacoustic electric-generator
technology,” in Proceedings of the Institution of Mechanical
Engineers, Part A: Journal of Power and Energy, 2018, doi:
10.1177/0957650918760627.
23. Y. Kondratenko, S. Serbin, V. Korobko, and O. Korobko,

“Optimisation of bi-directional pulse turbine for waste heat
utilization plant based on green IT paradigm,” Studies in

10.2478/pomr-2023-0064
QUANTUMNESS IN DIAGNOSTICS OF MARINE INTERNAL

COMBUSTION ENGINES AND OTHER SHIP POWER PLANT
MACHINES
Jerzy Girtler
Jacek Rudnicki *
* Corresponding author: jacekrud@pg.edu.pl (J.Rudnicki)
Abstract
The article provides proof that the diagnostics of marine internal combustion engines and other ship power plant
machines should take into account the randomness and unpredictability of certain events, such as wear, damage, the
variations of mechanical and thermal loads, etc., which take place during machine operation. In the article, the energy
E, like the other forms (methods) that it can be converted into (heat and work), is considered the random variable Et;
at time t, this variable has the mean value , which is the observed value of the statistic with an asymptotically
normal distribution , irrespective of the functional form of the random variable Et. A proof is given
that shows that the expected value estimated in the above way, considering the time t of the performance of task Z
by a marine internal combustion engine or other ship power plant machine, can be used to determine the machine’s
possible action (DM). When compared to the required action (DW) needed for task Z to be performed, this possible action
makes it possible to formulate an operating diagnosis concerning whether the engine or machine of concern is able
to perform task Z. It is assumed that an energy device of this type is able to perform a given task when the inequality
DM≥DW holds. Otherwise, when DM < DW, the device cannot perform the task for which it was adopted in the design
and manufacturing phase, which means that it is in the incapability state, although it still can be started and convert
energy into the form of heat or work..
Keywords: diagnostics, stochastic process, internal combustion engine, random variable
INTRODUCTION of a diagnosis about the state of SDN.

The process {B(t): t ≥ 0} is the process resulting from the
During the operation of any marine internal combustion use of SDG during the operation of the engine or machine
engine or ship power plant machine, irrespective of the applied performing a given task (SDN). This process is considered
diagnosis system (SDG), collecting the information needed to a long-term process, and it can but does not have to involve
formulate the diagnosis of the state of the engine or machine generating short-term diagnoses and/or formulating
that is the diagnosed system (SDN) requires the initiation prognoses or geneses. The course of this process has a
and continuation of the diagnostic process. This process [2], fundamental impact on the reliability of the diagnosis [12]–
[6], [10], [11] is a two-dimensional stochastic process {D(t, [14], [25], [38]. The process {C(ϑ): ϑ ≥ 0} is connected with
ϑ): t ≥ 0, ϑ ≥ 0}. It consists of the process {B(t): t ≥ 0}, which performing measurements of the current values of diagnostic
represents the SDG operation, and the process {C(ϑ): ϑ ≥ 0}, parameters (physical quantities such as temperatures,
which represents data collection by SDG and the formulation pressures, vibrations, etc.), with further diagnostic reasoning

performed in the short term, i.e. the time interval of the SDG accuracy and sensitivity of the measuring sensors and
operation (work) in which the diagnosis is obtained. transducers, the inaccuracy of the measuring devices,
The process {C(ϑ): ϑ ≥ 0} always consists of the following given by their inaccuracy class, and the stability and
realisations: diagnostic testing and diagnostic reasoning (of reliability of the measuring devices; or
the signal, measurement, symptom, structural, and operating • changes in the characteristics of the tested engine or
type) [4]. The output of the process {C(ϑ): ϑ ≥ 0} is a diagnosis, machine that take place during the measurement.
the reliability of which is highest when, during the testing The correct identification of the causes of the inaccuracy
process, SDG works reliably and the disturbances resulting of the performed measurement is necessary for the accurate
from environmental influences can be omitted due to the evaluation of the current inaccuracy of the characteristics of
sufficient resistance of SDG to these influences. the tested machine and that of the applied measuring devices,
In a general case, diagnostic reasoning is carried out using with the further correct selection of the proportions of these
the following types of reasoning in the following order: the inaccuracies. The difficulty in evaluating the inaccuracy
signal, measurement, symptom, structural, and operating of a measurement related to the properties of the applied
reasoning types [4]. Each of the above types of reasoning is measuring method and devices and that related to the current
characterised by the diagnostic uncertainty, resulting from characteristics of the diagnosed machine originates from the
the fact that diagnostic reasoning can be (and, as a rule, is) quantum nature of their changes, which leads to randomness
burdened with errors. As a result, diagnostic reasoning can and the unpredictability of events in the diagnostics of
lead to one of two possible mistakes to be made by the user marine internal combustion engines and other ship power
(diagnostician) of the engine or machine: plant machines. Therefore, this issue needs to be thoroughly
• a first-type mistake, which consists of diagnosing the considered.
engine or machine as being in the incapability state,
although it is still able to perform the given task, i.e.
it is in the state of capability; RANDOMNESS AND UNPREDICTABILITY
• a second-type mistake, which consists of diagnosing OF EVENTS IN DIAGNOSTICS OF MARINE
the engine or machine as being in the state of INTERNAL COMBUSTION ENGINES AND
capability, although it is already unable to perform
the given task, i.e. it is in the incapability state. OTHER SHIP POWER PLANT MACHINES
The above interpretation of the process {C(ϑ): ϑ ≥ 0}
shows that this process has values (states) that correspond In the diagnostics of not only marine internal combustion
to diagnostic tests and the above types of reasoning, and engines but also those used in cars, airplanes, etc., as well as
the durations of the diagnostic tests and the above types other machines of all types, it is advisable to stop using the
of reasoning are the execution times of these states. deterministic approach for the identification of the technical
A characteristic feature of this process is that certain state of these machines and focus on the probabilistic aspects
probabilities of occurrence can (and should) be attributed of their diagnostics [3], [6], [11], [17], [21], [24], [29], [30],
to its states, while the duration of each of these states is a [38]–[40]. The deterministic approach to the diagnostics of
random variable. marine engines and machines, as well as those used in other
During each diagnostic test, the measurements are made branches of technology, results from the traditional perception
with a certain accuracy, which depends on the applied of changes in their technical state, according to which it is
measuring methods and devices, as well as on the measurement believed that, in general, randomness and unpredictability
conditions and the experience and qualifications of the people can be omitted in technical diagnostics intended to assess
performing the test; all of these factors are possible sources the technical state of these types of devices. One of the main
of inaccuracy in the measurement. This inaccuracy results reasons for such an approach to the diagnostics of energy
from both the inaccuracy of the applied measuring methods devices was the fact that until the 20th century, a deterministic
and devices, and changes in the characteristics of the tested theory of the description of phenomena, events, and processes
engine or machine that take place during the measurement. was in force. This theory was developed by Pierre Simon de
The main causes of inaccuracy include the limited resolution Laplace, who assumed that similar laws of physics exist in
of the measuring devices (resulting from their sensitivity both the macroworld and microworld, and all changes take
threshold and the randomness of the examined phenomena) place according to these laws. They control the appearance
and errors such as the quantisation error, aperture error, and disappearance of each phenomenon, and the course of
and sampling time error when a digital signal is used in the all events (facts) and processes. According to this theory, the
measurement [27]. Hence, the diagnostic testing of a marine entire universe is totally deterministic on both the microscale
engine or other ship power plant machine is burdened with and macroscale. This vision of changes taking place in space
a certain measurement inaccuracy. This inaccuracy should and time was the basis for the development of science until the
be recognised well enough to determine its main cause, i.e. early 1920s – it was a basic methodological assumption made
whether it is mainly caused by [26], [27] by physicists. This view led to the foundation of mechanics,
• errors in the applied measuring methods and which now is referred to as classical, or non-quantum,
devices, which are known to depend mainly on the mechanics – in contrast to quantum mechanics, which was

developed later. It also led to the conviction that all the laws of observations well, due to the fact that gravitational fields
motion and any other changes can be expressed as differential existing in ordinary conditions are weak. However, according
equations that have unique solutions. This determinism can to the singularity theorems, the gravitational field is very
be found in the principles formulated by Isaac Newton to strong in two situations, at least: in the areas of black holes,
describe the laws of nature, in the partial differential equation and during and directly after the Big Bang [20]. Evidently,
proposed by Erwin Schrödinger (1926), the solution of which quantum effects cannot be neglected in these fields [17], [20],
is the wave function determining the quantum state of a [40]. We can expect that the classical theory of relativity
particle at an arbitrary time in a deterministic aspect, and in should finally collapse because of the above space-time
Albert Einstein’s equations describing the photoelectric effect singularities. Currently, research is in progress to develop the
and the relationship between energy, mass, and velocity [36], quantum gravity theory. Classical (non-quantum) mechanics
[37]. All these equations are not only deterministic but also was questioned because it assumed that atoms should collapse
time-reversible [22], [34]. However, despite the efforts of many to the state of infinite density. According to that theory, a
mathematicians, researchers have failed to prove the existence hot body should emit electromagnetic waves with the same
of unique solutions to differential equations, which was the intensity at all wave frequencies, which means that the total
inspiration for searching for a concept of the probabilistic energy emitted by this body is infinite. This conclusion is not
interpretation of reality. An example of such an approach true, and this was why Max Planck formulated a hypothesis
is the probabilistic interpretation of the wave function that electromagnetic waves cannot be emitted at an arbitrary
proposed as the solution to Schrödinger’s wave equation by rate but only as strictly defined portions, which he called
Max Born (in 1926), who could not accept the fact that this quanta (hence the name: quantum hypothesis).
function represents a ‘real’ electron wave, even though other It results from quantum mechanics that physical quantities
physicists accepted this equation as a tool for solving quantum such as energy or angular momentum can only change in
mechanics problems. In Max Born’s interpretation, the wave steps. Moreover, the quantities referred to as complementary
function Ψ is a product with complex numbers as values [5], have an important property: the simultaneous and accurate
[17], [19], such that is the probability of finding a particle measurement of their values is impossible. For instance, the
in a given area (point) in space. That means that there is no more accurate the position measurement of a microparticle
certainty about the exact position of an electron, but the (subatomic particle) is, the less accurate the measurement
probability that the electron is at a given point in space can of its momentum and, consequently, velocity will be. This
be calculated, provided that the wave function Ψ is known. is according to the Heisenberg’s uncertainty principle,
This interpretation corresponded to Niels Bohr’s opinion; he which defines the inaccuracy degree of the measurement
accepted partial and wave theoretical models of the existence of the above basic physical quantities (the position and
of particles. He also believed that we cannot predict an exact momentum of a particle, as well as energy and time). This
result of an empirical examination; in his opinion, we can only inaccuracy has nothing in common with the accuracy of
calculate the probability that the result of a given experiment the applied measuring methods and/or devices [5], [17], [19],
will take a given value and not another value. [20]. The uncertainty principle says that in the microworld,
However, the final blow to the deterministic theory of we cannot predict exactly the future position of a particle
Laplace came from the uncertainty principle formulated smaller than an atom, which is important, for instance, in
by Werner Heinsenberg (1926). Along with Max Planck’s controlling the stream of neutrons in a kinescope. Therefore,
quantum hypothesis (1900), which explained the nature it is understandable that the atom models proposed first by
of electromagnetic radiation generated by hot bodies, the Joseph John Thompson and then by Ernest Rutherford and
uncertainty principle became one of fundamental elements Bohr (although Bohr’s model quite precisely described the
(achievements) of quantum mechanics. Today, this theory structure of the hydrogen atom, as it is the simplest atom) were
is the basis for contemporary science and technology. It replaced by the quantum mechanical model of atom structure.
was developed in the 1920s by Werner Heinsenberg, Erwin In this model, the electrons in atoms do not move on specific
Schrödinger, and Paul Dirac, as well as by Wolfgang Pauli orbits; instead, they move in so-called orbitals, which are
and Niels Bohr. Additionally, Albert Einstein and Richard space regions around the nucleus in which the probability of
Feynman contributed to the development of this theory (the the existence of (finding) an electron at a given moment has
latter being the creator of nanotechnology). Its principles a precisely defined value. Following the proposal of Richard
explain, for instance, the functioning of transistors and Feynman, it was assumed that the particle does not move on
integrated circuits, i.e. most important components of one track but on all possible trajectories (permissible orbits)
electronic devices, without which modern diagnostics (not [1], [17], [20], [34]. These permissible orbits, called the orbitals
just technical) could not exist. These principles also apply in of electrons in atoms, are understood as space regions around
modern chemistry (quantum chemistry) [23], cryophysics the nucleus in which the electron can appear at a given time
(quantum liquid), and biology (medical diagnostics). with a certain probability [1], [19], [34].
Among the physical sciences, only the theory of gravity Transferring these conclusions to the macroscale
and cosmology has not been fully aligned with quantum research area, we can say that, according to Heisenberg’s
mechanics [20]. However, it may be expected that one day uncertainty principle, we cannot expect the same result when
this will happen. The general theory of relativity describes repeating any empirical research, regardless of whether it is

observational or an active experiment. Thus, the question of Another source of chaos can be inaccuracy in determining
how different the obtained results can be arises. The answer the model parameters. This fact is connected with the
is that the range of this difference depends on the adopted phenomenon of bifurcation (splitting between the real and
testing method, the accuracy of the measuring devices, the expected test results), which can be observed during the
current measurement conditions and their repeatability, the state identification of machines such as marine piston or
experience of the person conducting the test, the number of turbine engines, as well as positive displacement or rotary
measurements made, the duration time of the measurement, compressors, pumps, fans, electric motors, generators, etc.
etc. All of this means that when an empirical test is repeated, This phenomenon is an obstacle to obtaining a credible
in a particular experiment and in given conditions, different diagnosis, prognosis, or genesis.
results are always to be expected. This also means that The discovery of the principle of ambiguous causality
obtaining a specific test result is a random event, and the in science has led to the questioning of the earlier belief of
measured quantities should be considered random variables. unequivocal determinism (i.e. unequivocal effect resulting
Indeed, when the variability of the measurement results is from each cause) and the adoption of ambiguous determinism,
small, it can be neglected, but in each case, this decision i.e. determinism resulting from the probabilistic laws of
should be justified. Formulating a diagnosis first requires the quantum mechanics, which accepts the existence of choice
diagnostic test to be performed, as it is the first link in the (as a known rule).
diagnosis chain. The diagnostic test consists of measurements It results from the above considerations that when
made using a proper measuring device or the organoleptic constructing and using the diagnostic model of a marine
identification of the values of diagnostic parameters [3], engine or machine to formulate a complete diagnosis (i.e.
[4], [11], [17], [21], [24], [29], [30], [38]–[40]. The results of current diagnosis, genesis, and prognosis), the following laws
this test make it possible, using diagnostic reasoning, to (principles) should be taken into account:
formulate a relevant diagnosis (of the signal, measurement, • ambiguous causality, i.e. the existence of the
symptom, structural, and operating type) [4]. At each stage randomness of events (including events such as
of the diagnostic procedure, including diagnostic tests and machine state diagnosis), which indicates a need to
the subsequent types of diagnostic reasoning, the obtained accept at least ambiguous determinism,
results are burdened with the above-mentioned uncertainty • the uncertainty formulated by Heinsenberg,
and with errors caused by various disturbances. Therefore, • the existence of the general randomness of natural
the randomness of a diagnosis, prognosis, or genesis should phenomena resulting from their infinite complexity,
be considered an indispensable attribute. • the existence of deterministic chaos resulting from the
The quantum mechanics based on Heisenberg’s so-called sensitivity of models of empirical systems, in
uncertainty principle introduce unavoidable randomness particular internal combustion engines but also other
and unpredictability to science and engineering practice. machines (not only those installed in ship’s engine
A more general uncertainty than that defined by rooms), to their initial state,
Heisenberg’s principle is introduced by the phenomenon • the limited, as a rule, accuracy of the measuring
known as deterministic chaos [35]. This phenomenon can methods and devices, which leads to the limited
be observed when the tested model is a system of differential accuracy of the measurements made using these
equations, especially nonlinear equations of the 2nd, 3rd, and 4th methods and devices,
order. It is a well-known fact that the solution to a deterministic • the operating inaccuracy of the marine engine or ship
system of differential equations can take the form of very power plant machine that is the diagnosed system
complicated oscillations; the reason for this is not a large (SDN),
number of degrees of freedom, nor local disturbances, but the • the unreliability of the diagnosing systems (SDG)
increasing instability depending on the precision with which adopted to identify the technical state of a marine
the initial state is determined, which, in turn, depends on engine or ship power plant machine.
time-related initial conditions and time-dependent equation The measurements are associated with certain diagnostic
coefficients. Deterministic chaos is closely connected with procedures, during which some mistakes can be made as a
the occurrence of so-called attractors, which usually have the result of the following:
form of aperiodic trajectories that attract other trajectories • performing a diagnostic test in highly disturbed
from their environment [1], [38]. The detection of attractors conditions,
enables better prediction of the appearance of random events. • using an incorrect course of measurements
Therefore, recognising the fact that a given empirical system and incorrect error assessment (e.g. neglecting
develops chaotically may facilitate the study of its evolution. quantisation, aperture, and sampling time errors),
This means that chaos is not always a negative phenomenon. as a result of the application of measuring devices
Adding noise with random parameters to the non-disturbed with insufficient (inadequate) accuracy and/or the
empirical system can lead to statistical stabilisation or the omission of some measurements,
periodicity of the evolution of this system. This requires • incorrectly recording the results of measurements that
as new look at relations between deterministic diagnostic have been correctly performed and correctly signalled
models and those representing statistical and probabilistic by the measuring devices,
approaches.

• incorrectly interpreting the results of diagnostic tests case, further considerations concerning the diagnostics of
both during the diagnostic test and in the further machines will focus on demonstrating the suitability of the
steps of diagnostic reasoning, which is the result of generalised diagnostic system, which can be the operation
the inaccurate (incorrect) reading of the indicators of of a given marine engine or other ship power plant machine,
the measuring sensors (devices) and the application for determining the operational capability of these devices,
of inaccurate data processing algorithms, i.e. their ability to perform a given task in a given amount of
• incorrectly identifying the state of the engine or time and given operating conditions.
machine that is the diagnosed system (SDN), despite
correctly performing the measurements and obtaining
correct results from the diagnostic tests. THE ISSUE OF OPERATION OF MARINE
All of this means that in empirical diagnostic testing, INTERNAL COMBUSTION ENGINES AND
making use of certain measuring methods and devices, there OTHER SHIP POWER PLANT MACHINES IN
is a problem of measurement inaccuracy that results from
changes in the characteristics of the tested machine that take TERMS OF THEIR DIAGNOSTICS
place during the measurement and errors associated with
the use of certain measuring methods and devices [20], [21]. The operation of an arbitrary marine internal combustion
As a consequence, an indeterminacy appears that should be engine or ship power plant machine can be interpreted as
explained. In particular, the main cause of this indeterminacy of conversion of the energy E into the form of heat and/or
should be recognised, i.e. it should be determined whether work and its delivery to a receiver in a given time t (heat
it results from and work are forms, or – in other words – methods of
• a change in the characteristics of the engine or energy conversion) [15], [16], [31]. In this interpretation, the
machine that is the object of diagnostic testing that operation of each ship power plant machine, including the
took place during the measurement, or marine internal combustion engine, can be described (in an
• errors associated with the use of the given measuring evaluative approach) using a physical quantity with a given
methods and devices. numerical value and a unit of measure called the joule-second
Therefore, it is of high importance in this type of testing [joule×second].
to [28] Consequently, the operation of any marine internal
1. estimate the value of the operating uncertainty of the combustion engine or ship power plant machine can be
marine engine or other ship power plant machine that quantitatively determined using the physical quantity D (D =
is the object of the diagnostic test, E · t). This quantity contains information on how long the
2. estimate the value of the inaccuracy of the utilised energy E is or can be converted by a given engine or machine.
measuring technique (measuring method and devices), If we limit the analysis of the conversion of the energy E to
3. select adequate proportions between the accuracy only the form of work (L), then, taking into account the time
of the applied measuring technique and the current of this energy conversion, we can calculate the operation
inaccuracy of the tested object (a marine engine or of the engine or machine as DL = L · t. This type of data
machine). on the operation of a given engine or machine contains
It follows from the above that when using diagnostic information on how long the work L is or can be performed.
methods to determine the technical state of an engine or This information is as important as that about the power (N)
machine that is the diagnosed system (SDN) via an appropriate of a given marine internal propulsion engine or ship power
diagnosis system (SDG), it is difficult to obtain sufficiently plant machine, as it indicates how fast a given amount of
unambiguous answers to the following questions: work (L) can be done.
• What is the current structure of the tested machine With time, the operation of each marine internal
(SDN) and its resulting technical state ? combustion engine and ship power plant machine is becoming
• What were the causes that led to the present technical worse. Therefore, the issue that may be of a certain interest is
state of the machine ? the analysis and assessment of the operation of these devices,
• What will the specific properties of the SDN state be taking into consideration the above aspect.
during and after its future evolution ? The evaluation presented here of the operation of an
In this situation, formulating a specific diagnosis, arbitrary marine internal combustion engine or ship power
especially an operational diagnosis, requires the application plant machine has the following advantage: the descriptive
of mathematical statistics, probability calculus, and stochastic evaluation of its operation (the operation of the engine or
processes. Additionally, formulating the operational machine is good, acceptable, not very good, incorrect, bad,
diagnosis requires knowledge on the consequences of etc.) is replaced by an evaluation resulting from comparing
making a given decision that belongs to the set of possible the operation of a given engine or machine with another that
decisions. Nevertheless, a deterministic approach can be is used as a reference.
applied to determine the symptoms of the technical state The meaning of such an interpretation of the operation
of the machine; for instance, integral calculus can be used of a marine internal propulsion engine or ship power plant
to calculate the value of the machine’s operation. In this machine can be justified by the following reasoning: the

operation D (D = E · t) of an engine or machine (due to its
technical state) is better if more energy is delivered to the
receiver in a given amount of time (t). When the energy
transfer has the form of work (L), the operation (DL = L · t)
of the machine (due to its technical state) is better if more
work (L) is done by this machine in a given amount of time (t).
It is noteworthy that when the energy E is converted into
the combined form (method) of work L and heat Q, then, in
the evaluative approach, the following equivalence holds:
(1)
Fig. 1. Machine operation diagram: E − energy, E1 – energy attributed to time

which means that in this case, the value WE of the energy t1, E2 – energy attributed to time t2, t − time
E is equal to the sum of the value WL of the work L and the
value WQ of the heat Q, i.e. WE = WL + WQ.
In the case when the energy E is solely used to perform When this energy is converted in the time interval [t1, t2],
the work L (converted into the form of the work L), then, in the operation of the engine or machine can be interpreted
the evaluative approach, the following equivalence occurs: in general terms as follows [15], [16], [33]:
(2) (5)
which means that in this case, the value WE of the energy E where , D is the engine (machine) operation;
is equal to the value WL of the work L, i.e. WE = WL . E is the converted (obtained) energy that allows the realisation
Similarly, when the energy E is solely used for generating of the task in the time interval [t1, t2], and t is the time of
the heat Q, then, in the evaluative approach, the following conversion of the energy E.
equivalence occurs: Therefore, if we assume that , then
formula (5) can be written as (Fig. 1)
(3)
(6)
which means that in this case, the value WE of the energy E
is equal to the value WQ of the heat Q, i.e. WE = WQ. The use of formula (5) or formula (6) requires the geometric
The operation of a marine internal propulsion engine or application of a definite integral, and the following inequalities
other ship power plant machine can be considered using must be taken into account during the integration:
the following terms: required operation (DW) and possible
operation (DM) [16]. We can conclude that each engine or
machine is in the capability state, i.e. it can perform a given
task, when The integral given by formula (6) is the Riemann definite
integral [7], with the integration interval defined in this case
(4) as equal to [t1, t2] and the integrand . This
function is integrable in the Riemann sense in the above time
Otherwise, when DM < DW, we can conclude than the interval according to the following formula:
engine or machine is in the incapability state or partial
incapability state [15], [16], [32]. The capability of an engine or (7)
machine can be assessed after comparing the area of required
operation (DW) with that of possible operation (DM). This issue Hence, if we can determine the functional relation between
is discussed in [16]. the energy (E) and time (t) that characterises the operation
In a deterministic approach, the operation of a given of an engine or machine, i.e. the function E = f(t), and this
marine engine or other ship power plant machine can be function is continuous, for instance, in a given time interval
described using a general functional relationship describing [t1, t2], then, according to the second fundamental theorem of
the change in the energy E at an arbitrary time t of machine calculus (Newton-Leibniz theorem), we can write
operation. The operation of the marine engine or other ship
power plant machine analysed for E(t) = f(t) in a given time (8)
interval, e.g. [t1, t2], is shown as the area in Fig. 1.
The application of the Newton-Leibniz theorem is necessary
here because it enables the effective calculation of a definite
integral of any continuous function if an antiderivative of
this function is known.

In general, the functional relationship E = f(t) is not simple.
It is also possible that the antiderivative of the integrand
describing the relation between energy and time cannot be
defined by elementary functions. In that case, calculating
the definite integral using the Newton-Leibniz formula is
troublesome, and sometimes even impossible. The trouble
in this case is that determining the antiderivative requires
difficult transformations to be performed. In these cases,
similarly to the situation in which the integrand is given in
a tabular form, an approximate value of the operation of an
engine or machine can be determined as the value of the
definite integral calculated using the trapezoidal rule or the
Simpson method – the latter is considered more accurate. Fig. 2. Example of a stochastic process illustrating the relation E(t), where E
Taking into consideration the randomness and is a random variable: E − energy, E1 – energy assigned to time t1, E2 – energy
assigned to time t2, t − time as the process parameter, E[E(t)] – expected value
unpredictability of events that exists in operating practice of E, σ[E(t)] – standard deviation of E
and, as a consequence, in the diagnostics of marine internal
combustion engines and other ship power plant machines,
their operation can also be considered in the way described The estimation of the expected value of E(Et), which
above. However, in that case, the analysis and the resulting consists of its evaluation in the form of the arithmetic mean
evaluation of machine operation should be presented using a , is a point estimation. However, this estimation method
probabilistic approach, making use of the theory of stochastic does not provide opportunities for evaluating the accuracy
processes. A stochastic process is a random function with of the evaluation (estimation) of E(Et). Such an opportunity
time as a parameter. This approach to the evaluation of the is provided by interval estimation, which provides the
operation of marine engines and other ship power plant confidence interval [6], [8], [26].
machines results from the need to obtain information on The confidence interval of an unknown quantity E(Et)
the machine operation in the time interval between two is defined as the interval with random ends; it
arbitrary moments, e.g. [t0, tn], where time is the parameter contains the unknown value of E(Et) with a predetermined
of this process and not a random variable. In this case, to each probability β (the so-called confidence level) [8], [26].
time t within the given time interval [t0, tn)], we can assign
It is well known that the average calculated from
the state called the current state of the process, which is the
random variable Et; this variable has an expected value E(Et) formula (9) is the observed value of the statistic with
and a variance D2(Et) that depend on the current value of t.
an asymptotically normal distribution ,
It is not just the energy (E) that can be the variable in these
considerations; its conversion forms, i.e. work (L) or heat irrespective of the functional form of the random variable
(Q), can also be variables. Therefore, the stochastic process Et [6]. The quantities E(Et) and σ t represent, respectively, the
is a set of random variables Et for t ∈ [t0, tn], i.e. for t0 ≤ t ≤ tn. expected (average) value and the standard (mean) deviation
It is worth mentioning here that the expected value E(Et) of the energy E, which is a random variable at time t.
and variance D2(Et) of the random function {E(t): t ∈ [t0, tn]}
If the value of σ t, is known, then, making use of the
depend on t, i.e. the values of E(Et) and D2(Et) can be different
for different t values. They are not random functions of E(t), distribution of the statistic , we can
because E(Et) and D2(Et) are not random variables, but they are
calculate the confidence interval for an unknown expected
constant for a given value of t and a given set of realisations
of the random variable Et [8], [10]. value E(Et) from the following formula [8], [26]:
Examples of the dependence of E[E(t)], E[E(t)] +σ[E(t)],
and E[E(t)] – σ[E(t)] on the time t are shown in Fig. 2 [17]. In (10)
this figure, σ[E(t)] is the standard deviation of the random
variable E, which is calculated as the square root of the where yα is the standardised variable of the normal distribution
variance D2[E(t)]. corresponding to the confidence interval β = 1 − α.
Evaluating the expected value of E(Et) for each time t However, the value of σ t is usually unknown and should
requires the use of statistical inference, which consists of be estimated based on the obtained results of tests from the
the use of point or interval estimation. following formula:
It is known that the mean value can be calculated from
the following formula [8], [10]: (11)
(9) Then, assuming that the random variable Et has a normal

distribution , we can make use of the fact that

the random variable has the t-Student which is similar to that shown in Fig. 3.
After collecting a sufficient number of such realisations
distribution with k = n − 1 degrees of freedom. The (Fig. 3), we can calculate the characteristic parameters of the
stochastic process describing the relation E(t) shown in Fig. 2.
assumption about the normal distribution of the Another option is to use the model of the changes in the
random variable Et imposes no limitations in practice, as the operation of an internal combustion engine or other ship
power plant machine in the form of a homogeneous Poisson
statistic always has an asymptotically normal distribution process. This model enables the description of the decrease
and the convergence of this distribution to in the converted energy E in time t by an elementary portion
(quantum) e, which can be recorded by a measuring device
the normal distribution is very fast. This statistic can be used with a constant intensity λ > 0 (λ = idem).
for values n ≥ 4, i.e. always in practice [8]. Then, the course of the decrease in the energy E can be
Hence, the confidence interval can be calculated from the expressed as follows [2], [9], [15]–[17]:
following formula [8], [26]:
(13)
(12) A graphical interpretation of relation (13) is given in Fig. 4

for Ei (i = 1, …, 6). It shows that this process is discrete in
where is the coefficient of the t-Student distribution, states and continuous in time.
the values of which are such that .
In the proposed evaluation approach, studying the

operation of a marine internal combustion engine or other
ship power plant machine (pump, compressor) as a diagnostic
symptom of the technical state of this type of energy device
requires the collection of relevant statistics, which will make
it possible to determine the expected values of the energy
converted in these devices and further to attribute these values
to the individual times of operation of a given machine. Due
to the quantum nature of the measurement resulting from the
basic postulate of metrology, which is the assumption that the
sensitivity threshold 2ε > 0 [27], a sufficiently large number Fig. 4. Graphical interpretation of a sample realisation of the energy decrease
of diagnostic tests repeated over a relatively long period of of an engine or machine: E − energy, e − energy quantum by which the energy
E is decreased, which can be recorded by a measuring device, λ − intensity of
time during the operation of engines or machines can deliver the appearance of quanta (e) by which the energy E is decreased, as recorded
measurement results that will enable the description of the by the measuring device, t − time, E1 = Emax, E6 = Emin [16], [17]
energy in the form of the realisation of the process {E(t): ≥ 0},
which is discrete in states and continuous in time. When Another possible description of the decrease in the energy
studying the accumulation of the dissipated energy Er or a delivered by an engine or machine to the receiver, and the
decrease in the useful energy Eu for a given engine or machine resulting worsening of its operation, can have the form of a
due to its wear, we can obtain the realisation of this process, semi-Markov process, applied as a model of the operation of
Fig. 3. Interpretation of a) the accumulation of dissipated energy Er and b) the resulting decrease in the useful energy Eu of an internal combustion engine or other
ship power plant machine: e – portion (quantum) of energy by which the energies Er and Eu change, Erg – dissipated energy limit, Eug – useful energy limit [17]

this type of device [16]–[18]. a marine internal propulsion engine or ship power plant
machine can be examined by measuring the energy and the
time of this energy’s conversion; these results can then be
SUMMARY – REMARKS AND presented
CONCLUSIONS • as a number with a unit of measure called the joule-
second (formulas (5)–(8));
The diagnostics of marine internal combustion engines and • in a graphic form, as the area of operation (Figs. 1
other ship power plant machines should take into account and 2).
the randomness and unpredictability of the events that occur Despite the fact that it was formulated for marine internal
during their operation. Due to the operational practice of propulsion engines and ship power plant machines, the
power devices of this type and quantum mechanics, when interpretation of machine operation presented in this article
any empirical research is repeated, regardless of whether it can also be used to study the operation of other energy devices.
is an observation or active experiment, we cannot expect
the same results, but we can expect the same frequency of
occurrence of an individual result. This means that obtaining REFERENCES
a specific research result is a random event.
When performing diagnostic tests of marine internal 1. A. D. Aczel, Probability 1. Zysk i S-ka. Poznań 1997 (in
combustion engines or other ship power plant machines, Polish).
the principle of ambiguous causality should be taken
into account, which means that there is a need to accept 2. J. R. Benjamin and C. A. Cornell, Probability, Statistics,
ambiguous determinism, i.e. the determinism resulting from and Decision for Civil Engineers. WNT. Warsaw 1977 (in
the probabilistic laws of quantum mechanics, which accepts Polish).
the existence of choice (as a known rule).
Taking into account the results of tests performed in the 3. W. Batko, Z. Dąbrowski, Z. Engel, J. Kiciński, and S. Weyna,
phase of operation of marine internal combustion engines Modern Methods of Studying Vibroacoustic Processes.
or other ship power plant machines, a generalised diagnostic Radom Institute for Sustainable Technologies. Radom 2005
symptom has been proposed in the form of the operation of (in Polish).
any of the above-mentioned energy devices, assuming that
their energy values are becoming worse due to quantum 4. L. Będkowski, Elements of Technical Diagnostics. Military
energy dissipation. University of Technology. Warsaw 1992 (in Polish).
The operation of an arbitrary marine internal combustion
engine or other ship power plant machine is understood as 5. I. Białynicki-Birula, M. Cieplak, and J. Kamiński, Quantum
the conversion of the energy E in a given amount of time t Theory. Wave Mechanics. PWN. Warsaw 2001.
by these devices. This conversion process was compared to a
physical quantity that can be expressed with a numerical value 6. W. Cholewa and J. Kiciński, ‘Technical Diagnostics.
and a unit of measure called the joule-second [joule×second]. Inverse Diagnostic Models’. Gliwice, Silesian University
The operation understood in the above way becomes worse of Technology, 1977. (in Polish).
with time due to the increase in the wear of the engine or
machine. This means that the value of this operation in a given 7. K. M. Delventhal, A. Kissner, and M. Kulicki, Mathematics.
amount of time will decrease due to the decrease in the energy Compendium. Świat Książki, Warsaw 2005 (in Polish).
generated by the engine or machine. A suggestion was made
that in the case of the application of the theory of stochastic 8. S. Firkowicz, Statistical Evaluation of the Quality and
processes to analysing changes in the machine operation Reliability of Electron Lamps. WNT. Warsaw 1963 (in
understood in the above way, integral calculus can be used to Polish).
calculate machine operation parameters. A stochastic model
of the decrease in the useful energy generated by the engine 9. J. B. Gercbach and Ch. B. Kordonski, Reliability Models of
or machine in the form of a homogeneous Poisson process Technical Objects. WNT, Warsaw 1968 (in Polish).
was proposed to describe the range of the worsening of the
operation of the machine. A suggestion was also made to use 10. J. Girtler, ‘Controlling the Operation Process of Marine
for this purpose a model in the form of semi-Markov process, Internal Combustion Engines on the Basis of a Diagnostic
which is discrete in states and continuous in time. Decision Model. Monograph’, Gdynia, Notebook of Polish
When interpreted in the above way, the operation of an Naval Academy, 1989 (in Polish).
arbitrary marine internal propulsion engine or ship power
plant machine depends on its technical state and is jointly 11. J. Girtler, ‘Diagnostics as a condition for control of
characterised by the energy converted by this device and the operation of marine internal combustion engines’, Studia,
time of its conversion. vol. 28, Maritime School of Higher Education, Szczecin1997
In the version presented in this article, the operation of (in Polish).

12. J. Girtler, "Probabilistic measures of a diagnosis likelihood Warsaw 1984 (in Polish).
about the technical state of transport means’, Archives of
Transport, vol. 11, no. 3–4, pp. 33–42, 1999 (in Polish). 29. P. Puzdrowska, “Diagnostic Information Analysis of
Quickly Changing Temperature of Exhaust Gas from
13. J. Girtler, "Application of diagnostic reliability to decision- Marine Diesel Engine Part I Single Factor Analysis” Polish
making in the process of equipment operation’, 5th National Maritime Research, vol.28, no.4, 2021, pp.97-106, https://
Conference DIAG’2003 ‘Technical Diagnostics of Devices doi.org/10.2478/pomr-2021-0052
and Systems’, Warsaw – Ustroń 2003 (in Polish).
30. S. Radkowski, Vibroacoustic Diagnosis of Low-Energy
14. J. Girtler, ‘Probabilistic measures of the reliability of Damage. Institute for Sustainable Technologies. Warsaw-
diagnosis about the technical condition of machinery and Radom 2002 (in Polish).
other equipment’, XXXI National Symposium ‘Machinery
Diagnostics’, Węgierska Górka 2006 (in Polish). 31. J. Rudnicki, ‘Energy-time method for assessment of main
diesel engine operation,’ Journal of KONES Powertrain
15. J. Girtler, ‘Conception of valuation of combustion engine and Transport, vol. 14, no. 3, 2007.
operation”, Journal of KONES. Powertrain and Transport,
pp. 89–96, 2008. 32. J. Rudnicki, ‘Assessment of operation of energy system of
serial reliability structure on the example of ship main
16. J. Girtler, ‘Energy aspect of machinery diagnostics’, propulsion system,’ Journal of Polish CIMAC, vol. 3, no.
Diagnostyka, vol. 1, no. 45, pp. 149–156, 2008 (in Polish). 2, 2008.
17. J. Girtler, ‘Aspects of quantum machine diagnosis,’ XXXII 33. J. Rudnicki, ‘Quantitative assessment of operation of
National Symposium ‘Machinery Diagnostics’, Węgierska ship main diesel engine,’IV International Conference on
Górka 2012 (in Polish). Computational Methods in Marine Engineering, Lisbon
2011.
18. F. Grabski, ‘Theory of semi-Markov processes of operation
of technical objects’, Science Notebook of Naval College, 34. B. F. Schutz, An introduction to general relativity theory.
vol. 75A, Gdynia 1982 (in Polish). PWN. Warsaw 2002 (in Polish).
19. J. Gribbin, In Search of Schrödinger’s Cat. Zysk i S-ka. 35. M. Tempczyk, Chaos Theory for the Brave. PWN. Warsaw
Poznań 1997 (in Polish). 2002 (in Polish).
20. S. W. Hawking, A Brief History of Time. From the Big Bang 36. M. White and J. Gribbin, Einstein. A Life in Science. WNT.
to Black Holes. ALFA Publisher, Warsaw 1990 (in Polish). Warsaw 1995 (in Polish).
21. R. Michalski, Diagnostics of Working Machinery. Radom. 37. C. M. Wynn and A. W. Wiggins, The Five Biggest ideals in
Institute for Sustainable Technologies 2004 (in Polish). Science. Prószyński S-ka. Warsaw 1998 (in Polish).
22. M. Moskowits, Is science good. CiS Publisher. Warsaw 38. B. Żółtowski, Fundamentals of Machine Diagnostics.
1997 (in Polish). University of Technology and Agriculture. Bydgoszcz
1996 (in Polish).
23. R.F. Nalewajski, Fundamentals and Methods of Quantum
Chemistry. PWN. Warsaw 2001 (in Polish). 39. B. Żółtowski and C. Cempel, Eds., Engineering of
Machine Diagnostics. The Polish Association of Technical
24. S. Niziński, Elements of Diagnosis of Technical Objects. Diagnostics. Radom Institute for Sustainable Technologies.
University of Warmia and Mazury, Olsztyn 2001 (in Polish). Warsaw, Bydgoszcz 2004 (in Polish).
25. S. Pabis, Methodology and Methods of Empirical Sciences. 40. J. Kiciński, Ed., Modeling and Diagnostics of Mechanical,
PWN. Warsaw 1985 (in Polish). Aerodynamic and Magnetic in Power Turbine Units.
Institute of Fluid-Flow Machinery Polish Academy of
26. Z. Pawłowski, Mathematical Statistics. PWN. Warsaw 1976 Sciences. Gdańsk 2005 (in Polish).
(in Polish).
27. J. Piotrowski, Fundamentals of Metrology. WNT. Warsaw

2002 (in Polish).
28. Z. Polański, Planning Experiments in Technology. PWN.

10.2478/pomr-2023-0065
HUMAN RESOURCE MANAGEMENT DIGITALISATION

IN MULTIDISCIPLINARY SHIP DESIGN COMPANIES
Piotr Bilon
Wojciech Litwin
* Corresponding author: pb@wayman.software (Piotr Bilon)
Abstract
The digitalisation in the ship design sector is currently applied to the design process itself and is well defined, partially
standardised and practically implemented by both independent design companies and the design departments of
shipyards. The situation is similar in other sectors of engineering. However, the requirements for the digitalisation
of other processes in design and engineering companies have not previously been studied, and the limited financial
resources of ship design companies mean that there is a need for research on the digitalisation needs of multidisciplinary
ship design companies. The implementation of building information modelling (BIM) solutions is costly for design
companies and generates benefits mainly for shipyards and shipowners. The lack of benefits for design companies leads
to the hesitation of managers whenever digitalisation is considered; the scope and scale of the digitalisation, as well
as the selected software and BIM level, are defined by the shipyard requirements. The participation and support of
management in the digitalisation process is one of the key success factors; the expected benefits caused by digitalisation
for the organisation will increase the motivation of managers to invest limited resources in digitalisation. There are no
data that indicate the processes with a high potential for digitalisation and the scale of expected improvements in ship
design companies; therefore, research in this area was performed with a group of project managers from design and
engineering companies. The research focused on collecting the opinions and experiences of the managers related to the
manual management of resources and comparing the poll results with the conclusions from the enterprise resource
planning (ERP) system data analyses. The paper analyses if the digital automation of the resource management process
can lead to the substantial improvement of the operations of multi-project, multidisciplinary engineering ship design
companies.
Keywords: Ship design human resource management, digital automation, project portfolio, multidisciplinary engineering,
ship design, marine engineering
introduction most cases it leads to substantial efficiency improvements

[5]; however, it generates a number of risks [6], especially
All sectors of engineering are experiencing an intensive when the organisation is not ready for implementation [7].
digitalisation process [1]; this applies to the marine [2], The impact of digitalisation is clearly visible in the design
infrastructure, industrial, plant design, architecture system area and the design of data management systems [8].
engineering construction (AEC) and mechanical sectors Change itself, for engineers, is an integral part of the iteration
[3]. The implementation of software solutions for design and design process, which is usually turbulent due to the constant
engineering companies brings numerous benefits [4], and in modifications [9]. Considering a business point of view, there

should be a justification for a change, and usually there are
good economic reasons to make a change [10].
The direct costs of the software, implementation, training
and changes are covered by the design and engineering
companies, which can be the reason that in the short term,
digitalisation can be considered an unfavourable change.
Another challenge for the design and engineering sector is
the globalisation of services and distributed design networks.
There is the possibility of easy cooperation with designers
from different time zones and countries with lower cost
levels, which is not always is beneficial in terms of quality [11].
The subject of project planning and execution is strongly
connected with resource management [12], and for companies
facing aggressive competitors from low-cost countries, the
possibility of increasing the effectiveness of the organisation
and simultaneously reducing management costs by lowering
the amount of time required by the management process,
which often involves highly experienced and expensive
experts [13], [14], might be an interesting area to explore. It
is interesting from the scientific point of view to understand
the current solutions used by design and engineering
companies to manage resources and how much of the
project budget, in terms of time and financial resources, is
utilised for management [15]. Data-driven analyses could
provide answers that could be compared with the managers’
individual opinions, which can be collected using a poll. Fig. 1. Visualisation of the BIM model of the exhaust gas system
The conducted review of scholarly publications revealed of the cargo vessel
a lack of studies specifically targeting human resource
management processes in marine design firms; however, The greater the scale of the disturbances generated by
a prevalent recommendation among researchers in the changes, the more justified it will be to develop and implement
domain of project portfolio management and its human digital solutions that support the process of continuous
resource management sub-process is to conduct inquiries management [20]. The scale of the problem and the potential
within distinct engineering sectors, thereby catalysing for improvement need to be determined [21]. The reduction of
research endeavours within the marine design firm sector. the management time and a faster decision-making process
A good example of both multidisciplinary and diversified can generate benefits in the organisation and leverage the
engineering activity is the ship design sector. Based on efficiency of the engineering departments. In examining the
the matrix organisation structure, design and engineering need for innovative solutions in human resource management
companies are simultaneously performing several projects, for engineering and design projects, especially under
facing a large number of changes [16]. In these circumstances, conditions of change and uncertainty, the work by Litwin,
design and engineering resource management becomes Piątek, Leśniewski and Marszałkowski on the design and
a crucial foundation of competitiveness and the key to implementation of a hybrid propulsion system in a 50’ sail
business and technical success [17]. High-level digitisation catamaran provides a relevant example [22]. Their approach in
results in an increase in the design process’ economic effectively balancing multiple design criteria and navigating
efficiency by improving internal procedures and eliminating through complex engineering challenges underlines the
manual activities [18]. The implementation of available importance of adaptive and forward-thinking strategies in
software solutions allows the top management of engineering project management. This aligns closely with our discussion
design companies to focus on the decision-making process on the necessity of rapid and flexible planning in human
[19]. In Fig. 1, a graphical representation of the scope of the resource allocation, where the goal is to efficiently manage
multidisciplinary exhaust gas system conversion project resources in the face of evolving project requirements, much
developed by one of the Wayman software users is shown. like the multi-dimensional optimization challenges tackled
in their yacht design project. Also the innovative approach
detailed by Branowski, Zabłocki, Kruczewki and Walczak
in their study on the universal design of yachts for people
with disabilities demonstrates the necessity of adaptable
design solutions that cater to diverse user needs, mirroring
our discussion on the need for flexible and responsive
human resource planning in engineering projects, where

diverse and changing requirements must be met efficiently companies consume around 20% of the project budget, and
and effectively [23].The expected improvements might be 70 to 80% of the management budget is consumed by human
a good justification for investments and the implementation resource management activities. Moreover, the poll made it
of new solutions, and the solution selection process is an possible to understand managers’ opinions on the resource
interesting subject in itself [24]. Current global trends in management process and the tools used among the various
design and engineering companies lead to the conclusion that design and engineering sectors.
an increase in the subsidiary processes in the organisation
and IT tool integration should lead to an increase in efficiency
and collaboration; digitalisation is an obvious next step METHODS
for the construction and shipbuilding industry [20]. As
the implementation of the core business-related building The thesis, which boils down to the statement that the
information modelling (BIM) software generates the majority management of design and engineering resources is one of
of profits for the end customer, the construction company the key areas that needs to be digitised, leading to the need to
[25], the management personnel of the independent design eliminate manual management activities, has been verified in
and engineering companies strive to cut the costs and time a selected community of medium-level managers of selected
consumed by their own work [26]. Fig. 2 shows a visualisation design and engineering companies in Poland. As part of the
of the complex multidisciplinary scope of the Ro-Pax vessel research, using anonymous questionnaires, information
exhaust gas system developed by one of the Wayman users was collected from 1283 design and engineering company
using the BIM software. employees on how the problem of resource management is
perceived, and the content of the
databases containing information
on the labour intensity of manual
management activities was analysed.
The analysed databases were not
prepared for research and were
taken directly from the production
IT structure of the 20 engineering
companies operating in the ship
design, infrastructure, offshore,
infrastructure and oil and gas sectors.
The environment was not prepared in
order to precisely collect data related
to manual management.
Due to the standardisation of
the nomenclature, the numbering
of the activities performed and the
Fig. 2. Visualisation of the scope of a complex multidisciplinary exhaust gas possibility of importing data from
modification for a Ro-Pax vessel various software systems, it was achievable to combine
data from various design companies operating in various
Among the subsidiaries of the core business process, engineering sectors in order to compare and analyse the
those that require higher time and resource levels should be results. The first research step was to build an anonymous
prioritised, and the available resources should be allocated questionnaire that made it possible to obtain information
to the implementation of the most promising solutions from relevant to the analysis of the problem. The link to the
a business point of view [17]. In order to be able to properly research was sent to managers of design companies using
define the priorities and the scale of the potential positive one of the most popular ERP systems in Poland used for
impact on the organisation, it is necessary to understand the managing multi-sector engineering companies. Courtesy
current state of the art and the resource management activity of the software vendor of our own developed-in-house ERP
level [27]. This article describes results from the first iteration software, a customer communication channel was used to
of research related to the changes initiated by the continuous promote and deliver a request to complete the questionnaire.
resource management process in design and engineering Due to the need to protect the identities of users, the surveys
companies. The goal of the enterprise resource planning were anonymous. The accessibility of a large number of design
(ERP) system database analysis, performed in parallel to the company managers and the serious approach to the research
questionnaire research, was to collect and analyse the data supported by the software vendor, which distributed the
registered in the ERP used by the design and engineering survey through official software vendor communication
companies in order to understand how time and resources channels, justify the selection of the questionnaires as
are consumed in the resource management process. It is a research method. The survey research duration was 14
expected that project management activities in ship design months.

The first group of questions in the survey was used to and engineering sector. In the group mentioned above,
determine the characteristics of the enterprise in which the 70% (893/1283) of the respondents declared that resource
surveyed person is employed. Information was collected on management is performed manually using spreadsheets,
the scale of the company, the engineering sector, the types and the remaining managers also use the resource load
of industries, the applied solutions supporting management matrix, which is available in popular project planning and
and the share of the labour intensity devoted to manual management programs. Considering the impact of manual
management in the budget for project management and in the resource management on the consumption of the hourly
overall budget of the project. It is important to underline that budget dedicated to management activities, over 86% of
budgets were examined in both hourly and financial terms. respondents (1105/1283) believe that manual management
In order to initially verify the validity of the thesis, the activities consume 75% or more of the available time.
opinions of the respondents were examined to confirm According to the respondents, the cost of manual resource
the problem of a lack of digital resource management and management also remains high; 79% of survey participants
to determine its significance and the possible impact of (1010/1283) believe that on average, more than 75% of the costs
eliminating manual management. As part of the parallel allocated to project management go to manual management.
study, the databases of ERP system users were downloaded This is a value slightly lower (1010 vs 1105) than the number
and analysed in a way that allowed the results obtained of hours spent on the manual management of the hourly
by processing the data reported to the ERP system to be budget, but due to the fact that some activities related to
compared with the responses provided by the respondents. manual management can be delegated to staff with lower
The ERP system dedicated to engineering made it possible competencies, the values declared by the respondents seem to
to define task dictionaries, use a unified numbering of tasks be consistent. The analysis of the results of the study in terms
for all projects and assign tasks to appropriate groups, as of the impact of manual resource management on the budget
well as to use any attributes to group and process data and of the entire project, in terms of both the hourly and financial
information. Such attributes can distinguish to some extent budgets, indicates that the majority of respondents define the
the hours allocated to manual management activities and share of activities related to manual resource management
other management activities. to be at the level of at least 10%, in both hourly and financial
The ERP systems used in the design and engineering terms.
companies require employees to be assigned to precisely In the case of one company, it was possible to identify the
defined roles. Based on this feature, it is possible to identify date when the resource planning procedures were migrated
the users responsible for resource management activities. from manual management using spreadsheets to a resource
Typically, management tasks are performed by middle- matrix in the planning/project management software. The
level management experts (department managers, project organisation was performing, at the same time, a large
managers, discipline leaders, leading engineers) who can be amount of exhaust gas conversion multidisciplinary projects
generally classified as the employees who are permanently that, according to the company’s management, were similar
or temporarily responsible for teams of designers or for tasks in terms of the scope and level of design process disturbances.
performed by a group of people. It is a common practice that Before the implementation of the planning procedure, the
each employee is obliged to record the amount of time spent consumption was 2834 hours (2398 on design and 436
on certain tasks. This makes it possible to collect information on management); the scope of the work of the referenced
about who is doing what, compare the plan and the reality project is presented in Fig. 1. After the implementation of the
and easily calculate the cost of the work. In spite of the factmanagement,
planning procedure,
were similar thescope
in terms of the consumption onprocess
and level of design a similar project
disturbances. Before the
that there are local, national and corporation-wide standardsimplementation
was 2140 hours (1820 for design and 320 on management);
of the planning procedure, the consumption was 2834 hours (2398 on design and
436 on management); the scope of the work of the referenced project is presented in Fig. 1. After
for defining the tasks, there is no universal task-definingthe implementation
the graphical representation
of the planning procedure, theof the scope
consumption on aof the project
similar workwas of 2140
thishours
policy. Therefore, it is important to verify to which object(1820 project is presented in Fig. 2. The results are shown in Fig. 3.
for design and 320 on management); the graphical representation of the scope of the work of
this project is presented in Fig. 2. The results are shown in Fig. 3.
in each analysed database the time consumed by resource
management is registered.
Resources planning procedures used
Planning software used 1820 320
RESULTS
Descriptive analyses of the survey results were used for

the analysis of the questionnaire responses. Four hundred Manual in spread sheeet 2398 436
and thirty (430/1283) of the questionnaire respondents

are employed in the ship design sector. Ship design sector
0 500 1000 1500 2000 2500 3000
representatives are the largest group of respondents in Hours allocated
the research, and 1187 respondents confirmed that they
Design hours allocation Resources planning hours allocation
are employed by design and engineering companies. This
high conversion rate was achieved through cooperation Fig. 3. Comparison of the time consumption (in hours) before and after
Fig. 3. Comparison of the time consumption (in hours) before and after the implementation of the planning procedure
with a dedicated ERP software vendor for the design the implementation
Most of the respondents (1242/1283, orof93%)
the planning procedure
define the elimination of manual resource
management as an important aspect of improving the efficiency of a design company and state that
they develop several variants of future plans as part of manual management, while 75% of all
respondents (957/1283) state their interest in a software solution that supports the preparation of
future variants to support the resource management process in a design and engineering company.
POLISH
The combined MARITIME
results are RESEARCH,
presented in Table 1. No 4/2023 123
Table 1. ERP software user questionnaire results
1. Can you describe your company as a design and engineering multidisciplinary corporation?
No. Answer Res. %
1.1. Yes 1187 93%
1.2. No 96 7%
2. Please indicate your sector of engineering
2.1. Marine (ship design) 430 34%
Most of the respondents (1242/1283, or 93%) define the 7. How is the % of the project management financial budget
elimination of manual resource management as an important consumed on resources management?
aspect of improving the efficiency of a design company and 7.1. <5% 17 1%
state that they develop several variants of future plans as 7.2. <10% 697 54%
part of manual management, while 75% of all respondents 7.3. <15% 565 44%
(957/1283) state their interest in a software solution that 7.4. Other 4 0%
supports the preparation of future variants to support the
8. How big is the impact of resources manual management
resource management process in a design and engineering elimination on the efficiency of engineering company?
company. The combined results are presented in Table 1. 8.1. Critical 671 52%
Tab 1. ERP software user questionnaire results 8.2. Significant 571 45%
1. Can you describe your company as a design and engineering 8.3. Minor 32 2%
multidisciplinary corporation?
8.4. No impact at all 9 1%
No. Answer Res. %
9. Do you prepare several variants of resources management
1.1. Yes 1187 93% projection?
1.2. No 96 7% 9.1. Always 25 2%
2. Please indicate your sector of engineering 9.2. When requested and when I feel 654 51%
2.1. Marine (ship design) 430 34% it is needed
2.2. Infrastructure 356 28% 9.3. When requested 595 48%
2.3. AEC 120 9% 9.4. No, never, it is useless 9 1%
2.4. Plant design 257 20% 10. Would you be willing to use the software tool for automated
resources management even if it creates only variants for
2.5. Airspace 7 1% consideration?
2.6. Energy 113 9% 10.1. Yes 957 75%
3. Please indicate the methodology of resources management in your 10.2. No 258 20%
organization
10.3. I do not know 68 5%
3.1. Manual in spreadsheet 893 70%
3.2. In resources matrix in 387 30%
planning/project management For some companies, manual management is just a part
software of general department management; it might be registered
3.3. In dedicated automated 0 0% as a general management task, just like periodical project
resources management software meetings and even quality control activities. This means that
3.4. Other 3 0% in this first iteration of the research, the assignment of tasks
4. How big is the % of the hourly budget for project management to the appropriate category may be imprecise, due to the fact
consumed on resources management? that the users of the ERP system did not use attributes that
No Answer Res. % clearly define tasks for managers for tasks related to manual
management. This caused the need to manually review each
4.1. <25% 0 0%
stage of the selected projects and each task in order to manually
4.2. <50% 178 14%
assign the tasks and the hours allocated to each task to the
4.3. <75% 673 52% proper group. On the other hand, an unambiguous result is
4.4. Other 432 34% that 18–27% of the financial budget and 22–28% of the hourly
5. How big is the % of the financial budget for project management budget were used for activities related to the management
consumed on resources management? of project work and resources; this is applicable to virtually
5.1. <25% 0 0% all engineering sectors. Considering these numbers in the
5.2. <50% 273 21% context of the average margins on design and engineering
5.3. <75% 437 34% services should focus the attention of employees responsible
for business efficiency on this particular area of the company’s
5.4. Other 573 45%
operations.
6. How big is the % of the project hourly budget consumed on
resources management?
6.1. <5% 767 60%
DISCUSSION
6.2. <10% 389 30%
6.3. <15% 124 10% The subject of project portfolio selection and management
6.4. Other 3 0% is of scientific interest, and it is gaining momentum when it
comes to the number of publications on the topic. Currently,
researchers are exploring the area of decision support systems,
and the trend is to develop more management-oriented

solutions, as the usefulness of elegant and sophisticated business-related activities. The organisation described by
mathematical methods is often questioned [28]. Within the Hollister and Watkins was a retail company that faced
area of ship design and engineering management activities, resource overload and the incorrect allocation of resources.
including engineering resource management, real-life For the design and engineering sector, the scale of project
managers, who face a limited amount of available time and portfolio-related turbulence is much higher and turbulence
an urgent need to make decisions, always prefer the tools that occurs much more frequently compared to the example of
reduce the time required by a decision-making process over the retail company. The available data and the ability to
the complex methods that require the participation of an create variants of the future with various assumptions and
expert in order to be useful in real-life cases. The ship design priorities are obtainable but time-consuming. The scale of the
sector tends to migrate from the area where activities are difficulty and the involvement of managers in the resource
balanced among art, science and handcrafted approaches [29] management process in design and engineering companies
to less creative approaches, where designers use computational has not previously been a subject of research; therefore, our
methods and computer-based models. research had the goal of verifying the scale of the involvement
The same research shows, in its onion model, that currently, based on the available data and the opinions of the managers.
research related to project portfolio management is focused In this study, the results of research from two sources,
on coping with change and uncertainty [28]. Considering databases and the subjective opinions of the surveyed
the frequency of change in project portfolios, the ability to managers of design and engineering companies, were
predict the future is urgently required [30]. In terms of human compared. Currently, there is no data available from the
resource allocation in engineering, predicting the future is wider design and engineering sector that could be compared
naturally impossible, but the concept behind the research to the results achieved. At the time that this research
presented in this article is the real need for a quickly available began, there was no synthetic information in the analysed
list of variants of future human resource allocation created on customer databases that would allow one to determine the
the basis of parameters that are easy to define for managers. actual workload for activities related to manual resource
The goal in this respect is not to create the optimal solution management; however, it was possible to determine the total
as it would lead to the infinitive number of Pareto optimal costs of project management based on hard data, registered
solutions, but to provide input data for decision maker to working hours and the hourly rates of the reporting persons.
make a rational decision about the future plan of resources At the same time, the subjective assessments of managers
allocation. In addressing the complexities of human resource regarding the share of activities related to manual resource
allocation in engineering, particularly under the constraints management of the overall budget for project management,
of unpredictability and the need for rapid adaptation, which oscillated around 70% of the budget, indicate that
the principles outlined in Pawlusik, Szłapczyński, and resource management is an area of interest and probably has
Karczewski’s study on optimizing rig design for sailing yachts the potential for both direct and indirect cost reductions.
using an evolutionary multi-objective algorithm [31] offer Considering the justification for this research and the actions
valuable insights. Their approach in handling multiple, often leading to the digitalisation of human resource management
conflicting objectives through multi-objective optimization in maritime engineering design firms, it is vital to focus not
and multi-criteria decision making provides a pertinent just on the benefits generated for the organisation itself,
framework for our discussion on managing dynamic project which stem from the anticipated reduction in the amount
portfolios and resource allocation in engineering firms. Their of time dedicated to human resource management tasks.
methodology underscores the importance of balancing It is equally crucial to emphasise the broader advantages,
various factors, a concept that is integral to our analysis of which will include enhanced quality, more efficient budget
efficient and flexible human resource planning in the face of utilisation, the swifter development of value-driven projects
change and uncertainty. and a more effective execution of engineering services.
The current state of the art related to project portfolio As the database content does not allow the comparison
resource management is presented by Hollister and Watkins of data at the same level of detail, it is worth focusing on the
[32]. They specified seven causes of the resource overload results obtained from the surveyed managers. It is remarkable
problem: impact blindness, multiplayer effects, political that in design and engineering companies, according to the
logrolling, unfunded mandates, band-aid initiatives, cost majority of respondents, 70% of management work is still
myopia and finally initiative inertia. In the same article, the done manually. This result is naturally surprising, especially
authors provide a recipe for practically solving the project when the number of changes and modifications that affect
portfolio resource overload problem, and they share an project implementation plans in shipbuilding, and also in
interesting result of the research related to the scale of the other sectors of engineering, is considered. Reducing the
involvement of managers in initiatives that are non-productive amount of time spent on manual management decreases the
and not directly related to business, which reached 30% of direct costs of management, and most importantly, assuming
the overall time. In the described example, the company that a faster transfer of guidelines to the project team results
implemented a process for the manual review of plans and from the reduced management execution time, increases
selected the variant that allowed them to complete the top- efficiency by shortening the circulation time of the decision
priority projects and allocate more time and resources to loop.

CONCLUSIONS project company managers believe that 75% of the time
dedicated to project management activities is spent on human
The analysis of the results leads to the conclusion that resource management tasks. However, it was not possible
the development of solutions that support the management to reliably confirm these assertions, due to the lack of the
of human resources in engineering and design companies precise definition of activities belonging to the set of tasks
may address the essential needs of managers and increase related to human resource management. This implies that
the efficiency of a company’s work. Shortening the reaction in the future, there is a need to assign clear attributes to
time to changes can lead to the reduction of numerous risk tasks associated with human resource management. Another
elements, which can increase the company’s competitiveness significant conclusion is the confirmation of both a reduction
and the efficiency of engineers’ work. The effort to minimize in the hourly budget for project work and the budget allocated
the time needed for preparing new project in shipbuilding for project management after the introduction of a system that
is widely visible in the actions taken in other areas of ship supported management. This suggests that potential further
design and shipbuilding for example application of reverse benefits in this area could be achieved through automation
engineering in part design [33] and the use of AI algorithms for human resource management
The responses of the respondents show the importance in maritime engineering design firms.
of the management topic and the perceived need to spend
a large amount of time on management, with a simultaneous
careful approach to the topic of preparing many variants of REFERENCES
future projections, especially in terms of possible management
variants. The great interest of managers in the application for 1. A. Ghosh, D. J. Edwards, and M. R. Hosseini, “Patterns
the preparation of management variants also shows that the and trends in Internet of Things (IoT) research: future
topic is worth further exploration and is one of the areas of applications in the construction industry,” Engineering,
operation of an engineering design company that is probably Construction and Architectural Management, vol. 28, no.
difficult to learn using classic manual methods. 2. 2021. doi: 10.1108/ECAM-04-2020-0271.
The results of the research conducted lead to the conclusion
that the analysed area has the potential to be improved; the 2. R. Iwańkowicz, “A Multi-Case-Based Assembly
activities related to the management of resources can be Management Method for the Shipbuilding Industry,”
classified as problematic for managers, while the currently Polish Maritime Research, vol. 28, no. 2. 2021. doi: 10.2478/
used methods involve numerous risks and are ineffective. pomr-2021-0018.
The opinions of a selected group of managers indicate
that further exploration of data collected in ERP systems 3. V. Koscheyev, V. Rapgof, and V. Vinogradova, “Digital
is justified and may lead to a more reliable and transparent transformation of construction organizations,” in IOP
picture of the situation, provided that the collected data are Conference Series: Materials Science and Engineering, 2019.
properly prepared. Users should be properly motivated to doi: 10.1088/1757-899X/497/1/012010.
increase the amount of detail in their reporting to allow
a precise understanding of the time spent on management 4. J. Berlak, S. Hafner, and V. G. Kuppelwieser, “Digitalization’s
and the preparation of future variants. The binary data that impacts on productivity: a model-based approach and
have been analysed so far have not been prepared in a way evaluation in Germany’s building construction industry,”
that allows the proper identification of resource planning- Production Planning and Control, vol. 32, no. 4, 2021, doi:
related activities and the separation of the planning budget 10.1080/09537287.2020.1740815.
from other management activities; therefore, in the future,
an effort must be made to organise the database structure in 5. Q. K. Jahanger, J. Louis, C. Pestana, and D. Trejo, “Potential
order to make it possible to report resource planning hours positive impacts of digitalization of construction-phase
for separate objects. information management for project owners,” Journal of
The short-term prospect can lead to the wrong conclusions, Information Technology in Construction, vol. 26, 2021, doi:
but competitiveness, or even remaining active on the 10.36680/j.itcon.2021.001.
engineering market, is not possible without modern design
tools. Therefore, digitalisation for design and engineering 6. B. R. K. Mantha and B. García de Soto, “Assessment of the
companies is a part of reality that managers have to cope cybersecurity vulnerability of construction networks,”
with, face and cover the costs of. Engineering, Construction and Architectural Management,
The objective of the research described in this article has vol. 28, no. 10, 2021, doi: 10.1108/ECAM-06-2020-0400.
been partially achieved. Over a 14-month period of survey
research, 1283 responses were obtained; based on the content 7. S. Demirkesen and A. Tezel, “Investigating major challenges
analysis of databases from ship design companies, these for industry 4.0 adoption among construction companies,”
responses confirmed that over 20% of a project’s hourly Engineering, Construction and Architectural Management,
budget is allocated to management-related activities. The vol. 29, no. 3, 2022, doi: 10.1108/ECAM-12-2020-1059.
survey research further revealed that 86% of the surveyed

8. Jan Koeleman, Maria João Ribeirinho, David Rockhill, 18. M. Fusko, Ľ. Dulina, P. Bubeník, M. Bučková, M. Kasajová,
Erik Sjödin, and Gernot Strube, “Decoding digital and R. Svitek, “THE IMPORTANCE OF DIGITIZATION
transformation in construction,” McKinsey & Company, AND INNOVATIONS FOR SMALL AND MEDIUM-
no. August, 2019. SIZED ENTERPRISES,” Proceedings of CBU in Economics
and Business, vol. 2, 2021, doi: 10.12955/peb.v2.252.
9. Z. Wang and J. Liu, “A Seven-Dimensional Building
Information Model for the Improvement of Construction 19. M. Winfield, “Construction 4.0 and ISO 19650: A panacea
Efficiency,” Advances in Civil Engineering, vol. 2020, 2020, for the digital revolution?,” Proceedings of Institution of
doi: 10.1155/2020/8842475. Civil Engineers: Management, Procurement and Law, vol.
173, no. 4, 2020, doi: 10.1680/jmapl.19.00051.
10. S. Parusheva, “Digitalization and Digital Transformation in
Construction – Benefits and Challenges,” Proceedings of the 20. M. Eigner, “Engineering 4.0—Foundations of the
International Conference dedicated to the 50th anniversary Digitalization of Engineering,” in System Lifecycle
of the Department of Informatics, University of Economics, Management, 2021. doi: 10.1007/978-3-658-33874-9_3.
Varna, 2019.
21. M. Müller and L. Jedličková, “Several Notes on the
11. L. Zheng, W. Lu, K. Chen, K. W. Chau, and Y. Niu, “Benefit Existential Hermeneutic Phenomenology for Project
sharing for BIM implementation: Tackling the moral Management and Possibilities of Its Extension by Other
hazard dilemma in inter-firm cooperation,” International Existential Concepts: Case Study From the Research Project
Journal of Project Management, vol. 35, no. 3, 2017, doi: Team,” Project Management Journal, vol. 51, no. 4, pp.
10.1016/j.ijproman.2017.01.006. 452–463, Aug. 2020, doi: 10.1177/8756972820910280.
12. M. Rauf, Z. Guan, L. Yue, Z. Guo, J. Mumtaz, and S. 22. W. Litwin, D. Piątek, W. Leśniewski, and K. Marszałkowski,
Ullah, “Integrated Planning and Scheduling of Multiple “50’ Sail Catamaran with Hybrid Propulsion, Design,
Manufacturing Projects under Resource Constraints Theoretical and Experimental Studies,” Polish Maritime
Using Raccoon Family Optimization Algorithm,” IEEE Research, vol. 29, no. 2, 2022, doi: 10.2478/pomr-2022-0012.
Access, vol. 8, pp. 151279–151295, 2020, doi: 10.1109/
ACCESS.2020.2971650. 23. B. Branowski, M. Zabłocki, P. Kurczewski, and A. Walczak,
“Selected Issues in Universal Design of Yachts for People
13. R. Moser, G. Narayanamurthy, K. Kusaba, and G. Kaiser, with Disabilities,” Polish Maritime Research, vol. 28, no.
“Performance of low-cost country sourcing projects – 3, 2021, doi: 10.2478/pomr-2021-0030.
Conceptual model & empirical analysis,” Int J Prod Econ,
vol. 204, pp. 30–43, Oct. 2018, doi: 10.1016/j.ijpe.2018.07.011. 24. L. Venz, A. Pundt, and S. Sonnentag, “What matters for
work engagement? A diary study on resources and the
14. H. Abbasianjahromi and S. Hosseini, “A risk-cost benefits of selective optimization with compensation for
optimization model for selecting human resources in state work engagement,” J Organ Behav, vol. 39, no. 1, 2018,
construction projects,” SN Appl Sci, vol. 1, no. 11, 2019, doi: 10.1002/job.2207.
doi: 10.1007/s42452-019-1570-5.
25. J. Bin Yang and H. Y. Chou, “Subjective benefit evaluation
15. B. Ashuri and M. Tavakolan, “Shuffled Frog-Leaping Model model for immature BIM-enabled stakeholders,” Autom
for Solving Time-Cost-Resource Optimization Problems Constr, vol. 106, 2019, doi: 10.1016/j.autcon.2019.102908.
in Construction Project Planning,” Journal of Computing
in Civil Engineering, vol. 29, no. 1, Jan. 2015, doi: 10.1061/ 26. L. Rihar, T. Žužek, and J. Kušar, “How to successfully
(asce)cp.1943-5487.0000315. introduce concurrent engineering into new product
development?,” Concurr Eng Res Appl, vol. 29, no. 2, pp.
16. E. Acar, G. Bayrak, Y. Jung, I. Lee, P. Ramu, and S. S. 87–101, Jun. 2021, doi: 10.1177/1063293X20967929.
Ravichandran, “Modeling, analysis, and optimization
under uncertainties: a review,” Structural and 27. N. Bibi, Z. Anwar, and A. Ahsan, “Comparison of search-
Multidisciplinary Optimization, vol. 64, no. 5. 2021. doi: based software engineering algorithms for resource
10.1007/s00158-021-03026-7. allocation optimization,” Journal of Intelligent Systems,
vol. 2015, pp. 629–642, 2015, doi: 10.1515/jisys-2015-0016.
17. S. Saeed, M. A. Khan, and R. Ahmad, Business strategies
and approaches for effective engineering management. 2013. 28. L. K. Hansen and P. Svejvig, “Seven Decades of Project
doi: 10.4018/978-1-4666-3658-3. Portfolio Management Research (1950–2019) and
Perspectives for the Future,” Project Management
Journal, vol. 53, no. 3, pp. 277–294, 2022, doi:
10.1177/87569728221089537.

29. A. Karczewski and J. Kozak, “A Generative Approach
to Hull Design for a Small Watercraft,” Polish Maritime
Research, vol. 30, no. 1, pp. 4–12, 2023, doi: 10.2478/
pomr-2023-0001.
30. P. F. Drucker, E. Dyson, C. Handy, P. Saffo, and P. M. Senge,

“Looking ahead: implications of the present.,” in Harvard
business review, 1997.
31. M. Pawłusik, R. Szłapczyński, and A. Karczewski,

“Optimising Rig Design for Sailing Yachts with Evolutionary
Multi-Objective Algorithm,” Polish Maritime Research, vol.
27, no. 4, 2020, doi: 10.2478/pomr-2020-0064.
32. R. Hollister and M. D. Watkins, “Too many projects,”

Harvard Business Review, vol. 2018, no. September-October.
2018.
33. M. Deja, M. Dobrzyński, and M. Rymkiewicz, “Application

of Reverse Engineering Technology in Part Design for
Shipbuilding Industry,” Polish Maritime Research, vol.
26, no. 2, 2019, doi: 10.2478/pomr-2019-0032.

10.2478/pomr-2023-0066
STRATEGIES FOR DEVELOPING LOGISTICS CENTRES:

TECHNOLOGICAL TRENDS AND POLICY IMPLICATIONS
Minh Duc Nguyen

Faculty of Economics, Vietnam Maritime University, Haiphong, Viet Nam
Ko Tae Yeon
Heesung Electronics Vietnam Limited Company, Viet Nam
Vietnam Maritime University, Haiphong, Viet Nam
Krzysztof Rudzki
Gdynia Maritime University, Faculty of Marine Engineering, Gdynia, Poland
Hoang Phuong Nguyen
Academy of Politics Region II, Ho Chi Minh city, Viet Nam
Nguyen Dang Khoa Pham*
PATET Research Group, Ho Chi Minh City University of Transport, Ho Chi Minh city, Viet Nam
* Corresponding author: khoapnd@ut.edu.vn (Nguyen Dang Khoa Pham)
Abstract
Logistics centres are currently performing a key function in the development of countries through their ability to regulate
goods, markets, and transport. This is shown by the infrastructure, cost, goods flow, and quality of logistical services
provided by these centres. Nevertheless, in developing nations or regions with antiquated logistics infrastructure,
conventional logistics centres seem to struggle to manage the volume of commodities passing through them, resulting
in persistent congestion and an unsteady flow of goods inside these facilities. This issue poses a challenge to the
progress of any nation. The emergence of new technology offers a potential avenue to solve the problems inherent in
traditional logistics centres. Most prominently, four technologies (the Internet of Things (IoT), Blockchain, Big Data
and Cloud computing) are widely applied in traditional logistics centres. This work has conducted a thorough analysis
and evaluation of these new technologies in relation to their respective functions and roles inside a logistics centre.
Furthermore, this work proposes difficulties in applying new technologies to logistics centres related to issues such
as science, energy, cost, or staff qualifications. Finally, future development directions, related to expanding policies
in technological applications, or combining each country’s policies for the logistics industry, are carefully discussed.
Keywords: Logistics centers, logistics infrastructure, logistical services, smart technologies
within businesses. The term ‘Industry 4.0’ frequently refers

introduction to a wide range of digitalisation methods, approaches, and
technologies because it largely focuses on applications in
An important topic for the growth of the world’s economy the industrial environment. Because of this, the application
and contemporary society is digitalisation. Continual sectors range from enhancing material flows to buildings,
expansion and long-term welfare are cited as essential drivers manufacturing, and product development. These programmes
of successful digitalisation efforts. Therefore, government should help industrial organisations achieve their objectives,
programmes that seek to develop innovation capacity, increase including higher competitiveness based on open processes,
productivity, decrease costs, increase revenues, improve more agility, improved adaption, and increased flexibility.
preparation for the digital era, and strengthen competitive The initial commercial practices and academic studies on
advantages are regularly used to promote digitalisation goals disruptive technologies are still very much relevant in today’s

society. Disruptive technologies were also the foundation aims to improve things like object traceability and the ability
of modern business models like Amazon and Flipkart, to understand their current location or status; (ii) Internet-
since they outperform incumbent technologies, in terms of oriented, which seeks to improve network protocols like the
productivity, efficiency, and convenience [1]. The Internet of Internet Protocol, which is seen as the network technology to
Things (IoT), for instance, completely replaced warehouse connect smart objects all over the world; and (iii) Semantic
and inventory management through a precise combination of oriented, which centres on issues of meaning and context
supply hubs, transportation, and customer handling system, [5]. In addition to IoT, other new technologies have also
which was a boost for e-commerce industries. As a result, begun to become more prominent during this period,
IoT could offer more individualised, responsive, and novel such as Big Data, Blockchain, AI, etc. Each technology has
or unconventional customer service, in addition to decreased different functions that are combined in an industry, thereby
operational costs [2]. The Internet of Things is anticipated to helping the industry develop strongly in all respects [6][7].
play a significant role in the logistics sector in the near future. The logistics industry is no exception to this development;
It is also evident that many objects and items have already hence, the use of 4.0 technologies has transformed logistics
begun to carry or tag bar codes, RFID tags, and sensors, into a strategic tool for gaining a competitive edge, shifting
bringing geospatial data and allowing tracking of a variety its perception from a simple financial burden to a valuable
of goods and merchandise through a single supply chain asset [8]. Table 1 presents a comprehensive compilation
from any location [3]. Primarily, there are three schools of of definitions of Industry 4.0 technologies that have been
thought with respect to the IoT [4]: (i) Things oriented, which specifically implemented in industrial environments.
Tab. 1. Definition of outstanding 4.0 technologies [9]
4.0 Technology Definition Ref
Internet of Things (IoT) The implementation of sensors and devices that are networked through wireless networks and [10][11][12][13]
Internet-based interaction with the objective of enhancing the value of products and processes.
Blockchain The digital platform facilitates the safe storage and distribution of information across a collective of [14][15][16]
users via the creation of time-stamped, tamper-proof, and indefinitely lasting records. The system
comprises decentralised ledgers that store transactions as data blocks, which are interconnected
by a cryptographic pointer. The aforementioned system exhibits attributes such as distributed
consensus, enhanced security measures, traceability, verification, and transparency of information.
Cloud Computing The online service provides users with the ability to do rapid and streamlined calculations, without [17][18][10]
the need for establishing a tangible infrastructure. This technology facilitates the provision of
computer resources, including networks, servers, storage, applications, and services, with the
ability to access a network that is readily available, easily accessible, and beneficial. The outcome
is a more economically efficient and expeditious resolution with regard to operational platforms,
software, and infrastructures.
Big Data The effective facilitation of decision-making processes is achieved via the management of [19][20][5]
a substantial amount of data, defined by its high volume, rapid velocity, and diversity. This is
accomplished by using cutting-edge analytic approaches that are creative in nature.
The logistics industry is showing an increasing dependence of logistics systems. As a result, the whole manufacturing
on new technologies. Therefore, in order to fully capitalise on sector, including its logistics, is transitioning to a paradigm
the potential of this industry, it is imperative to develop the that is more adaptable and agile, making room for Industry
‘Logistics 4.0’ initiative, which aims to maximise the utilisation 4.0. According to Khatib et al. [22], the heart of Industry
of cutting-edge technologies and implement innovative 4.0 consists of four primary enabling technologies that will
advancements in the logistics field [21]. Governments should increase the adaptability of manufacturing and distribution
aggressively push its ‘Logistics Centre 4.0’ strategy, starting processes: robotics, Big Data, wireless networking, and
with the logistics centre, which serves as the beating heart inexpensive sensors. The interdependencies between the
various 4.0 technologies are depicted in Fig. 1.

Fig. 1. Dependency between 4.0 technologies [22]
In the technologies mentioned above, IoT was acknowledged literature, logistics centre management is a crucial component
as one of the most significant fields of future technology and it of the supply chain that has received increased attention,
is one of the key information and communication technologies with the aim of meeting the rising freight demand and the
(ICT). The IoT is quickly gaining ground in the context of increasingly high standards for logistics services [29]. In the
contemporary wireless telecommunications, particularly last few decades, the warehouse and logistics centre business
with the rapid development of wireless communication has experienced a remarkable development in major trading
technologies [23][24][25]. From an initial emphasis on cities. Planners and social scientists have expressed concerns
machine-to-machine communication and applications in over the social and environmental impacts associated
the ‘ubiquitous aggregation’ of data, the concept of IoT is with this increase. Researchers have shown that transport
continually changing. In other words, the IoT has generated infrastructure and links to supply chain partners are two of
vast amounts of data and various mathematical analytic the most important factors in luring developers to an area
methods can be used to continuously investigate the intricate [30][31]. Although the advantages of the effects and the role of
links between the transactions represented by this data. digitalisation in logistics are still not properly recognised, it
Without a doubt, IoT would be crucial to the deployment cannot be denied that digitalisation will be an essential step in
of smart logistics [26][27], which would fundamentally alter the field of logistics [32][33]; this is reflected in transportation
the design of the logistics system and the logistical operation and warehouse activities [34]. As a typical example, in the
mode. These changes are very important in determining past, real-time tracking of vehicles was performed using
a company’s competitiveness because logistics costs are GPS systems. Then, with the advent of 4.0 technology and
seen as a significant component of overall production costs. blockchain, real-time goods tracking systems using blockchain
Numerous businesses are thinking about how to operate their technology have been applied by many companies. From
warehouses more cost-effectively and efficiently, especially there, digital transformation frameworks are continuously
in light of recent advancements in supply chain and logistics being updated, based on emerging logistics activities. The
technology but there are also many businesses that are hesitant research by Junge et al. [35] presented a framework for digital
about applying smart technologies to a previously traditional transformation in logistics, as demonstrated in Fig. 2.
infrastructure [28]. According to the public and academic

Fig. 2. The Logistics Digital Transformation Framework [35]
The next problem for the necessary development of 4.0 consistently high in underdeveloped nations. Precisely for
technologies is the e-commerce industry. The quantity of this reason, the Warehouse Management System was born to
products that need to be handled in a logistics centre has resolve the backlog of manual operations at warehouses and
grown over the past ten years, to meet the development of the logistics centres. Warehouse Management Systems (WMS)
e-commerce industry, making warehouse operations more evolved as a result of the emergence of increasingly complex
complex. As a result, older manual methods are no longer tools and algorithms to run warehouses effectively in the
sufficient or practicable to handle this enormous amount 2000s [45]. A WMS has an information system that combines
of activity [36]-[37]. From there, freight and logistics firms software programmes to keep track of, regulate, control, and
may encounter large delays between arrival and departure, manage inventory levels, and optimise warehousing choices.
as a result of poor management of the logistics centre [38]. Order processing, order release, and master data are the three
Extra fees (fines for late transportation and additional main WMS functions. In addition to the above functions, the
expenditure for keeping the tractor-trailer driver on overtime) additional capabilities of WMS include receiving (inbound),
and orders being delivered late are the direct result of this putting things away, and warehouse control [46]. However,
[39]. Long lines of idle vehicles contribute to pollution [40]- WMS stopped meeting the needs of traditional logistics
[41], thus it is best to set up transport and storage facilities centres. For the e-commerce industry, specific characteristics
with standardised loading and warehousing efficiency. The related to the volume and quantity of goods still make it
primary objective of transportation and logistics centres difficult for WMS to operate processing and storage.
is to minimise the overall costs associated with product By recognising the importance of technologies in
transportation, while simultaneously increasing storage applying the management of logistics centres to a smart
capacity through the implementation of short-term strategies environment, this work analyses and makes judgements
and the use of technology. These elements play a crucial role in of Industry 4.0 applications in the intelligent management
the efficient delivery of goods and enhancing transportation of flogistics centres. Applications that reduce the number of
and logistics systems [42]-[43]. Moreover, the absence of pointless procedures can lighten the load in a logistics centre.
cohesion in the managerial procedures within logistics centres Efficiency and operational productivity are then improved.
leads to frequent instances of time wastage and mistakes in These applications also pave the way for future studies in the
the tasks performed by personnel teams. Additionally, one field of smart logistics, where systems are combined with
company’s barcode system will differ from their suppliers’, cutting-edge technology. However, there will be some notable
leading to inconsistencies in the preservation of information obstacles, such as the cost of deploying the technology, the
about the items’ features. This has long been a problem in way in which the location of the logistics centre affects the
conventional logistics centres, where it slows down the arrival environment, and the policies for applying these technologies;
of transporting vehicles and drives up waiting times and these will be of interest in future research.
other expenses [44]. Because of this, the cost of logistics is

Literature Review transportation terminal, it is a crucial link in the multimodal
transportation chain, and is a structure that serves as a hub
for transportation activities among various modes of
Logistics centre definition transportation. With the birth and evolution of the phrase
‘supply chain management’, the logistics centre concept
According to the European Economic Interest Group has also altered and developed, much like the concept of
[47], a logistics centre is “the hub of a given area where all logistics itself [60][61]. Nonetheless, it is most apparent that
activities connected to transport, logistics, and products a logistics centre needs to have two main functions: shipping
distribution both for national and international transit are and warehousing. These are regarded as the two most crucial
carried out, on a commercial basis, by multiple operators”. elements in setting up a basic logistics centre.
Facilities for storage, managing, clearing, reassembling,
disassembling, inspecting quality, offering lodging, and Smart logistics centre
providing social services are all found in logistics centres.
Logistical activities are moving from urban to rural areas All facets of social life are being significantly impacted
and using environmentally friendly modes of transportation, by the Industrial Revolution 4.0. Whether they want it or
like electric vehicles. In the maritime field, the development not, people are impacted by the revolution every single
of green solutions for ships and ports or green logistic day. Automation, labour-saving production, lightning-fast
centres is being considered as a priority, aiming to reduce product speed, and consistent quality are at the core of the
carbon emissions and mitigate climate change [48]–[50]. fourth industrial revolution. One of the primary concerns
In industrialised nations, logistics centres are crucial for is the use of automation as a means to effectively adapt to
sustainability and competitiveness, but they are also helpful the strategic change that affects the national economy [62].
for regional development in poor nations [51]. For logistics centres, the 4.0 revolution is bringing a lot of
On the other hand, Uyanik et al. [52] asserted that an technology to support smarter and more flexible operations.
essential component of development strategy was the use of New technologies are seen as a great support to meet the
logistics centres, which were initially developed in Europe automation needs of logistics centres. Applying technology
in the 1960s and were first observed in the US during the to logistics centres will help effectively manage the quantity,
industrial revolution [53]. If such a centre was established in status, and flow of goods, in terms of time and cost.
conjunction with combined and intermodal transport types, The concept of ‘smart logistics’ has emerged in recent
there would be innumerable advantages to doing so, including times, whereby advanced information technology serves as
lower prices, reduced traffic congestion, lower environmental the fundamental basis for its implementation [63]-[64]. By
pollution levels, etc. However, the term ‘logistics centre’ was processing information from all facets of logistics in real-
never defined in the literature. According to Higgins et al. time and thoroughly evaluating it, contemporary integrated
[54] and Rimiené et al. [55], a number of terms implied a logistics systems can be intelligently implemented. End-
logistics centre, including distribution centre, freight village, to-end visibility, improved transportation, warehousing,
dry port, inland port, load centre, logistics node, gateway, distribution processing, information services, and other
central warehouse, freight/transport terminal, transport aspects of smart logistics could all result in time and
node, logistics platform, logistics depot, and distribution money savings. Additionally, it might be able to lessen the
park. Furthermore, Erkayman et al. [56] also published a environmental damage that logistics causes. However, there
concept for a logistics centre. National and international are still difficult problems that must be solved before smart
locations, known as logistics centres, are where various logistics can be implemented. One of the outstanding issues
operators conduct all logistics-related activities on a foris the application of technology and connected activities
profit basis, including shipping and forwarding, product in a logistics centre into a complete chain of activities, to
distribution, material handling, storage, and other related overcome additional time and cost. In order to fulfil the needs
transactions (such as banking and insurance). To carry out of product storage, surveillance, safety, fire and explosion
the aforementioned tasks, a logistics centre must be furnished detection, and more, a smart logistics centre needs IoT
with the necessary public amenities. These centres must be equipment, such as IoT stacking shelves and an IoT inspection,
located near connections to highways, railways, airports, as well as a monitoring system. The same degree of automation
and seaports, as well as being located outside residential and network connectivity needs to be applied to machines
areas. Lastly, it has to be run by a single public or private and tools. It is obvious that IoT will play a pivotal role in the
organisation [46][57]. success of these modern logistics centres. These applications
From the perspective of multimodal transport, according of IoT allow for the construction of a network-based cyber-
to Smail et al. [58], a logistics centre is a type of output logistics system that can be managed by humans. The
point structure of the supply chain that includes stages like incorporation of intelligence, automation, and automated
warehousing, distribution, and the provision of value-added choices of technology (IoT), elements of the supply chain,
services, and storage. According to Kaynak et al. [59], and logistics 4.0 are other important topics to explore in
the logistics centre is a hub that combines various modes the Fourth Industrial Revolution [65][66]. A modern smart
of transportation, almost performing like a multimodal logistics centre, with intelligent goods arrangement and

inventory management systems, is shown in Fig. 3. This is the To meet the needs related to the intelligence of a logistics
result of the combination of many Industry 4.0 and Logistics centre, the study by Yavas et al. [68] focused their research on
4.0 ideas, including driverless cars, digital connections, and the transformation of logistics centres in industrial revolution
information security, with the core functions of a logistics 4.0 and identified key criteria for logistics centres in the new
centre. industrial era. The strategy was to look at the interactions
between the operations of traditional logistics centres and
then suggest a new framework for them. The four primary
operations of logistics centres (handling management,
information management, transport management, and
warehouse management) were the basis for the twelve criteria
for logistics centre 4.0 presented in this work. These criteria
were acknowledged as the operational criteria at the logistics
centre stated above and were linked to the four traditional
criteria in the proposed framework [69], as illustrated in Fig. 4.
The operation of a smart logistics centre is based on the use
of the latest technology, such as Big Data and IoT, to increase
its operational efficiency. Through a decision support function
based on logistics data, it also aids in the development of a
managerial logistics operation plan. According to Cho [70],
Logistics Centre 4.0 is based on IoT technologies. The data is
collected and analysed using Big Data technology, the product
is stored and transported based on the knowledge obtained
through Artificial Intelligence, and a smart logistics centre
system performs tasks automatically using robots. As the
variety of items that need to be processed in a warehouse
Fig. 3. Operations of a smart Logistics Centre [67] has grown over the past decade, conventional and manual
techniques for warehouse management have proved to be
unable to manage them.
Fig. 4. A conceptual architecture for the 4.0 logistics hub [68]

This has resulted in a rise in the use of information Additionally, embedded systems are also utilised, in
technology to facilitate warehouse operations. The WMS conjunction with network communication technology, to
has evolved as more sophisticated tools and algorithms for achieve these objectives [26][74][75]. A sensor is apparatus
managing warehouses have been made available since the that identifies and reacts to a certain kind of stimulus
turn of the 2000s [45][71]. With the advancement of both originating from the surrounding physical milieu. The input
technology and the marketplace, however, the manufacturing might include a range of environmental phenomena, such
of products has shifted from make-to-stock (MTS) to make- as heat, light, motion, moisture, pressure, or other factors.
to-order (MTO), leading to a dramatic rise in the number of Alternatively, the data might be communicated by electronic
commodities. As a result, new forms of warehouse technology means over a network, enabling remote access for reading
are being integrated with existing warehouse management or further processing [76]. The classification of sensor types
systems. In this article, a smart logistics centre is defined is presented in Table 2.
as having four 4.0 technologies: IoT, Blockchain, Big Data, Connecting and collecting massive amounts of data from
and Cloud Computing. These technologies are applied for sensor systems in logistics hubs is made possible by Internet of
solving problems in logistics centres. In the next section, we Things applications. This data can come from many different
analyse and review some applications that have been, and sources, such as product volume, temperature, humidity, shelf
are being, applied at smart logistics centres. In addition, location, etc. According to Uckelmann et al. [85], the IoT is
applicable policies, development possibilities, and future dedicated to connecting the physical world to the virtual
research directions will also be mentioned in this work. internet and its primary drivers of development are object
self-identification, information sharing, and interactive
processing. Machine-to-machine and human-to-machine
Application of 4.0 technologies interactions are made possible by IoT applications [86]. The
in logistics centres technological components of the IoT have been extensively
discussed in Miorandi et al. [87], Mishra et al. [88], Ng et al.
The secret to creating a smart logistics centre is to use [89], Whitmore et al. [90], and Li et al. [91]. Despite the fact
cutting-edge technologies to their fullest potential. Of these, that several types of smart logistics exist due to different
IoT technology has emerged as a technology with outstanding priorities, all of them rely on the use of ICTs. A new working
data collection potential, helping managers to have a more concept of smart logistics is centred on the movement of
holistic view of a logistics centre [72]-[73]. To facilitate data goods, based on cutting-edge technology and intelligent
communication, exchange, and control among objects using management. Fig. 5 shows the concept map of a smart logistics
distinct identifiers, a range of information sensing technologies centre with four technologies in the current industry 4.0 era:
are employed. These technologies encompass RFID, wireless IoT, Big Data, Blockchain, and Cloud computing [92].
sensor networks (WSN), and machine-to-machine systems.
Tab. 2. Classification of sensors [77]
Type of sensor Function Ref

GPS sensors The objective of this sensor is to determine the location of various components and to accurately detect [78][79][80]
and communicate the operational duration of equipment in the logistics centre.
Strain sensors Quantify the instantaneous deformation experienced by structural components in a timely manner. [81]
Accelerometer sensors This particular sensor is capable of detecting changes in gravitational acceleration, enabling [78][79]
the measurement of tilt, vibration, and acceleration.
Barometric sensors Barometric pressure sensors have been employed in many electronic devices such as smartphones, smart [78][79][82]
watches, and drones, to monitor air pressure readings and changes in altitude.
Wind-sensor, rain-sensor The objective is to observe and measure the velocity of wind, as well as the amount of precipitation. [83]
Fibre optic sensor The automation of operations might be achieved by the activation of the reader for RFID and GPS detectors. [84]
Laser sensor The objective is to ascertain the duration required for the production process of the equipment. [78]
Fig. 5. Intelligent logistics: a conceptual roadmap [92]

Internet of things (IoT) Öner et al. [100] took a case study into account, for the
application of RFID technology in the wool yarn sector. RFID
One potentially useful technology that might be included is intended for the handling process, including receiving,
in a standard logistics system is the Internet of Things (IoT). picking, and shipping of semi-finished goods, as well as
Utilising Auto-ID methods, such as Barcode and Radio tracking work-in-progress, inventories, and stock levels. In
Frequency Identification (RFID), the IoT can accurately order to accomplish this, an architectural framework for an
identify a wide range of things. The IoT would gather and RFID-based information system for the wool yarn sector
record data in real-time from a wide variety of things, was created and a cost-benefit analysis was carried out to
allowing for real-time visibility and traceability [93]. Data determine whether the new system was cost-effective or
collected in real-time might be used for complex tasks like not. Additionally, a risk analysis for RFID investments was
logistics route planning. The logistics sector has recently performed. In the same study direction, a WMS was created
looked at wearable gadgets that combine IoT technology [94]. by Tejesh et al. [101], based on RFID wireless communication
Logistics platforms have included state-of-the-art technology technology. The IoT-based warehouse inventory management
to allow for automated tracking of commodities and improved system is designed to track the products that are linked to tags
capabilities for logistics personnel [95]-[96]. With the help of and provide product information and their associated time
IoT technology, a traditional logistics centre has the ability stamps for additional verification. A server called Raspberry
to connect devices and units, and manage those units within Pi would monitor and update all data. The entire system
a network. Fig. 6 shows the connectivity and management provides an archetype to match the material and information
capabilities of IoT in a logistics centre. flow. The website is designed for ease of use and an interface
Fig. 6. IoT application in logistics centre management [97]

for the user to track the products in mind. Comparing the
developed system to the already-used warehouse inventory
In order to promote intelligent logistics for Industry 4.0, management systems, it was significantly more affordable
Lee et al. [98] suggested an IoT-based warehouse management and operated dynamically.
system with an enhanced data analysis methodology, based on A logistics centre’s facilities were connected to a cloud
computational intelligence approaches. Data acquired from centre, gateways, fog devices, edge devices, and sensors
a case firm demonstrated that the suggested IoT-based WMS in Lin et al.’s [102] investigation of the deployment of an
might increase warehouse productivity, picking accuracy, and intelligent computing system. This work established an integer
efficiency, while also being resilient to order unpredictability. programming model for deploying gateways, fog devices,
The authors also discovered that employing RFID might and edge devices in their respective potential sites, so that
increase efficiency in order picking by warehouses, pickup the total installation cost was minimised under constraints
time, and inventory accuracy. A smart WMS framework was of maximal demand capacity, maximal latency time,
announced by Hamdy and colleagues [99]. The warehouse coverage, and maximal capacity of devices. The locations
manager could perform more real-time management and of the cloud centres and sensors were determined based on
monitoring of the activities because of this solution. The the factory layout. The system’s deployment was simulated
adoption of WMS and IoT in warehouses was reviewed. The using a mathematical programming model, which chose
building components and levels of the IoT were also shown, the locations of the gateways, fog devices, and edge devices
carefully. in the logistics centre, to minimise the overall installation

cost while maintaining the system’s maximum capacity for that would otherwise be impossible to reach, driving the
demand, latency, coverage, and device capacity. Two new warehouse closer to its full potential. In the study by Xie et
paths in the study of the vehicle routing problem (VRP) in al. [108], the researchers conducted a survey of companies,
transportation management (under the umbrella of smart looking at how Big Data is being used in the management
logistics) have emerged, because of intelligent technologies of logistics, and used logistics hubs as a case study for using
like IoT and ICT. First, studies on VRP began to focus on time series models to predict cargo load capacity. The results
multi-objective models and enhanced intelligent algorithms showed how smart logistics built on Big Data may improve
for handling dynamic optimisation problems. Data-driven logistics in many ways, including efficiency, cost, and user
models and dynamic models with various objectives have experience. The authors also concluded that the leadership
drawn more attention from academics, in terms of model and making of decisions, customer relationship maintenance,
types, because they address real-time data updating and and resource allocation of logistics firms would all greatly
coordination amongst numerous transportation agents. benefit from the judicious use of Big Data technologies.
Second, some researchers have shown that the use of Big Wang et al. [109] also conducted research on the topic of
Data and geospatial positioning technology enables smart locating logistics facilities through Big Data analysis. This
logistics to perform activities like visualisation, prediction, issue was stated in the form of a nonlinear mixed-integer
control, and decision-making in VRPs [103]. programme. The simulation analysed the effect of varying
For the purpose of organising fresh integrations of demand, distribution centre operating costs, international
intelligent food logistics systems, Li et al. [104] presented shipping, and client count on the optimal placement of
a linear approach. The costs of overall production, inventory, distribution centres produced in random, massive datasets.
and transportation were minimised, while average food The experimental data showed that the model provided
quality was increased. Then, a fresh approach was created was practical and stable. This case study demonstrated the
to resolve it by fusing the fuzzy logic method with constraint- practical use of Big Data in designing a distribution network
based two-stage heuristics. One case study and 185 randomly by evaluating different potential network layouts.
generated cases, with up to 100 stores and 12 periods, were Big data could be especially useful in inventory management,
used to evaluate the methodology. The case study’s calculation as mentioned in the study by Wang et al. [110], where it could
results showed that the suggested model and method could aid the development of cutting-edge inventory optimisation
resolve a real situation involving 40 merchants and 7 systems, the forecasting of future inventory requirements,
periods. The authors’ method could give decision-makers the meeting of fluctuating customer demands, the cutting of
a selection of Pareto solutions and assist them in selecting inventory costs, the attainment of a more complete picture of
a preferred alternative. Moreover, Zhang [105] suggested stock levels, the improvement of inventory flow and storage,
a path decision approach based on an intelligent algorithm, and the reduction of safety stock. Big Data provided additional
merging the Cyber-Physical System’s characteristics with the information about the logistics hubs that support certain
existing logistical system. The equipment layer’s connectivity industries. An advanced data mining strategy was presented
architecture and data processing utilised the IoT and cloud by Vieira et al. [111], for an automobile sector firm based
platform data storage technologies, which were based on the on their analysis of proof of concept Big Data in a logistics
Cyber-Physical System’s logistics path decision model. The centre. Due to the dearth of pre-existing methods, the most
impacts of using ant colony, simulated annealing, and genetic cutting-edge one was employed. To better identify relevant
algorithms on logistic path optimisation were thoroughly data to assist decision-making, the suggested strategy focused
examined. It was determined that the ant colony algorithm on goals that were user-driven. Another shared objective was
had the best path optimisation impact in solving the logistics to facilitate communication and consensus during decision-
path decision, by comparing the shortest transport route and making. In order to ensure that the proper replacement parts
convergence rate under the three algorithm decisions. were available for the right equipment at the right time and in
the right amount, Zheng et al. [112] proposed an intelligent
Big Data system for managing inventory that makes use of cutting-
edge technologies, such as the IoT and Big Data Analytics.
Data obtained from the market, such as consumer The Singapore Economic Development Board anticipated
preferences and the experiences of logistics users, can be that this approach would benefit the whole of the Singapore
processed and analysed with the use of big data gathering semiconductor sector in the future. The interactions between
technologies, allowing logistics firms to improve the quality providers and consumers should be investigated further to
of their services in response to client needs. Additionally, find ways to improve openness, adaptability, and satisfaction.
logistics firms can improve their overall competitiveness by Furthermore, Wahab et al. [113] set out to learn what
collecting data about their target markets, such as logistical variables in Malaysia’s warehousing industry were slowing
costs, basic pricing, and marketplace assets [106][107]. In down the use of Big Data analytics. The theoretical
order to improve supply chain management and logistics underpinning was the TOE model (technology-organisation-
centre efficiency, Big Data analytics sparked a revolution environment). Partial least squares structural equation
in inventory monitoring, forecasting, and management. By modelling was used to evaluate survey responses from 110
analysing massive amounts of data, Big Data provided insights logistics firms. The empirical findings indicated that the level

of adoption of Big Data analytics was influenced by relative A paradigm for the combination of blockchain with the IoT
advantage, technical infrastructure, absorptive capacity, was presented for a logistics centre by Aleksieva et al. [116].
and government backing, but that industry rivalry had little This technology, which made use of smart contacts on the
impact on noticeable gains. The results of this research should blockchain, might be used for the logistics of cross-docking
make it easier for warehouses to use Big Data analytics in warehouses and shipping. The model showed the ability to
the most effective ways possible. Big Data analytics are more operate effectively when it was possible to classify items in
likely to be adopted by warehouses that place an emphasis a logistics centre very well, thereby increasing operational
on operational excellence, ICT infrastructure, and the efficiency and reducing waiting times. Blockchain logistics
integration of new technologies. ‘apps’ for the optimum placement of intelligent transport
The inclusion of Big Data in data collection in the warehouse logistics centres were created by Chen et al. [117], to make
is a very feasible option, making data reception more passive. use of the blockchain system’s simplicity and IoT input
In addition, the combination of Big Data and IoT is also a good devices. This strategy was used to monitor where goods
technological solution for helping the logistics centre become were in the supply chain at any given time. Based on the
automated. This was also a premise for conducting transport results of the experiments, it was clear that the optimum
flow management outside the logistics centre. The combination location technique is superior to conventional approaches,
of managing the flow of goods and motor vehicles in and out of in terms of the amount of computing required, the precision
a logistics centre would greatly help in reducing transportation of the locations it produces, the total cost, and the ease with
costs, warehousing costs, and waiting costs. which warehouse locations might be determined. Applying
an intelligent logistics system built on the IoT and blockchain
Blockchain technology helps businesses get a clearer picture of their stock
levels and shipping progress in real-time. Thereby, it ensures
For IoT or Big Data, the application of these technologies in the assets and capital turnover of the enterprise.
logistics centres mean exploiting and processing information It is impossible not to mention the traceability of
to achieve optimal goals, as well as reducing costs. However, blockchain technology as, with this ability, many applications
the major drawback of the above applications is their have been launched, e.g. product traceability. In order to
transparency, as well as the ability to protect data, and this is reliably record airplane parts and improve traceability
what blockchain can do. By using examples and frameworks, data (with organisation-wide consensus and verification),
Ahmad and colleagues [114] explored how blockchain Ho and colleagues [118] suggested a blockchain-based
technology might revolutionise port logistics and operations. approach that was constructed using Hyperledger Composer
In addition, researchers designed permissioned architectures and Hyperledger Fabric. Blockchain was also used to keep
to draw attention to the numerous elements, participants, track of inventories at distribution facilities, cutting down
and deployment options of port logistics services, in order to on inefficiencies in both time and money. Kurdi et al. [119]
automate these processes. The results showed that blockchain analysed data from a sample size of 303 respondents, using
technology could render it impossible for theft to happen, regression and hypothesis testing with ANOVA, to apply
with documents linked to data management and storage, fleet a descriptive, exploratory, causal, and analytical design. One
management, trade paperwork, as-set and crew approval, and particularly noteworthy finding was the effect that blockchain
tracking shipments. This made transactions go more smoothly and smart inventory systems had on the efficiency of supply
and built trust between authorities, organisations, and other chains and logistics hubs. Future studies should expand on
players in the logistics centre transportation environment. the number of sectors and building types studied by using
With RFID tags, the supply chain process was described from the same amount of organised research. On the other hand,
the raw materials to the consumer. Each step was recorded in Lakshmi et al. [120] used QR codes and blockchain technology
the blockchain to improve transparency, in which the activities to create a system for trustworthy distribution and open
at the logistics centre would be recorded, see Fig. 7.
Fig. 7. The use of blockchain technology inside a supply chain framework [115]

inventory management. Distributors, retailers, suppliers, and data transfer. Sharma et al. [127] illustrated research on how
manufacturers might all be linked via blockchain technology, to leverage cloud computing to improve retail warehouse
with every transaction between them being permanently distribution and supply chain management through
recorded. The use of QR codes helps the efficient management Microsoft Azure technology. Microsoft Azure is utilised in
of this stock. Faster feedback loops mean fewer mistakes in retailing, shipping, and warehousing, as well as supply chain
inventory records and more reliable data for making well- management. The utilisation of this technology helps cost
informed decisions at review intervals. savings and eases administration, with more adaptability and
better oversight. These are just some of the benefits of using
Cloud computing this kind of technology in retail centre logistics. Gupta et al.
[128] also confirmed that a suitable application created by
The term ‘cloud computing’ refers to the delivery of cloud computing may greatly facilitate the simplification and
computing resources and functions through the internet. The automation of logistics centre management. Sivakumar et al.
key features of cloud computing include on-demand service [129] investigated the storage facilities at Chennai Harbour.
delivery, widespread network access, shared resource pooling, The warehouse management system for the Chennai Port
scalability, and use monitoring and control. Since a large Trust is hosted in the cloud, so that numerous people may
number of users might share identical assets, cloud-based use it from off-site locations. According to the findings of this
platforms automatically monitor and measure the usage of study, warehouse operations benefit from cloud computing,
resources for each user [121]. This allows users to make as little which increases productivity and streamlines the workflow.
or as much use of the system’s capabilities as they see fit [122]. Furthermore, Barreto et al. [130] noted that electronic contacts
The most-mentioned advantages of cloud computing is that it with customers, trade partners, and carriers may be handled
reduces risks in the supply chain and limits the generation of by integrating warehouse administration and transportation
waste. Supply chain risk might be mitigated and robustness administration using cloud computing technologies.
improved in the same manner as cloud computing in logistics
increases agility (by boosting speed, scalability, and visibility).
In a poll by Accenture [123], 52% of supply chain executives Policy implication for logistics
claimed that cloud computing has helped them improve centre development
resilience. The executives also claimed a 26% improvement
in the precision of demand projections as a result of using Smart logistics centres powered by the IoT, Big Data,
cloud computing. The effectiveness of supply networks in Blockchain, and Cloud Computing would be safer, more
reducing waste and their long-term viability are under accurate, and more efficient. Additionally, the warehousing
increasing examination. This was a major topic of discussion procedure would be expedited, and resources would be utilised
during COP26, held in Glasgow in 2021. Overall, 48% of to their full potential. In a centralised warehouse management
supply chain executives polled by Accenture in 2021 [123] system, decentralised decision-making was possible using IoT.
said that they had reduced waste because of cloud computing. The hurdles for IoT-based smart warehousing still lie in the
Companies might use cloud computing to highlight waste selection of a storage allocation strategy and the optimisation
and inefficiency in the supply chain and save costs, allowing of indoor routing. The technical limitations of RFID, the IoT’s
them to make adjustments to reduce their waste. The cloud limited technological capability, IoT standardisation issues,
might assist businesses in reorganising their supply chains to IoT data acquisition and processing issues, and IoT security
improve efficiency, refine logistics and transport routes, and and privacy concerns, are all obstacles to IoT-based smart
maximise resource use, all of which contribute to a smaller logistics. IoT is unsuitable for sophisticated applications in
carbon footprint [124]. The application of cloud computing logistics due to its constrained computing power and data
to each stage in the supply chain was an inevitable trend of processing capabilities. In transportation, IoT has a limited
businesses moving towards smart logistics. In the research by impact on cargo load optimisation and vehicle selection. IoT
Jiang [125], the researcher took advantage of cloud computing technology is unable to easily tackle the complicated problems
to provide a strategy for determining the best geographical of storage allocation and container/truck loading [131]. IoT
and transitable parameters for an international e-commerce could not be used to achieve the agile WMS [132]. Without
logistics distribution hub. When micro-influences were taken the aid of an intelligent algorithm, IoT technology could not
into account, this model performed well. Transportation provide decision-making in truck route optimisation for
distances were stated to be reasonable, ranging from 3.5-7.5 the delivery of perishable goods [133]. IoT technology could
km, when using this approach. Based on this technology, gather large amounts of information about delivery resources
Zhang [126] also presented a two-layer unloading system for and requirements [134][135], but it could not address how to
a railway logistics centre using cloud-edge communication improve the scheduling and use of delivery resources [20].
technology. The simulation findings demonstrated that the It is not sufficient to address the issues of resource waste and
discharging technique described in this research reduced excessive cost in last-mile delivery.
the total time cost of unloading by as much as 40%. This Logistics become more difficult during the handling,
technology’s use also has the potential to expedite inter- palletising, and transporting of custom or limited-edition
device communication and enhance the efficacy of railway products [136]. The occurrence of eventualities modifies

the demand for specific products quickly and calls for
adjustments to the warehouse floor plan. Transportation
scheduling changes at short notice also presents another
difficulty for traditional logistics. In light of the transit stock
being transported along the supply chain (and its scarcity,
needs, and trends), the issues in current logistics centre on
these factors [137][138], including spikes in the demand
for otherwise stable products [139]. The desire to deploy
Industry 4.0, according to Pereira et al. [140], involves a
variety of technological hurdles with significant effects on
many aspects of today’s manufacturing industry. Therefore,
before implementation begins, it is crucial to define a plan
for all the actors engaged in the entire value chain and to
come to an agreement on security-related concerns, as well Fig. 8. Links between logistics strategy and other industries [145]
as the appropriate architecture. Furthermore, a lot of scholars
claim that putting Industry 4.0 into practice is a difficult
task that would probably take ten or more years to complete. The use of 4.0 technology also requires ties with other fields
Adopting this new manufacturing method involves several and sectors within a nation. Furthermore, only a small portion
factors and presents a variety of obstacles and challenges, of 4.0 technology is used in logistics centre management.
including social, political, and economic issues, in addition Here, the focus is on building a nation’s digital infrastructure
to technical, economic, scientific, and energy hurdles. The and smart infrastructure. Although there are many questions
gathering, processing, and presentation of manufacturing about 4.0 technology’s ability to safeguard information, its
process data were three new, demanding activities that practical applicability, and the expense of doing so, this is
they should be able to test out using specialised Industry thought to be one of its most risky. Theft, or the disclosure of
4.0 technology [141]. Industry 4.0 has the potential to bring user-data kept in databases, might compromise user privacy
about significant changes in a number of areas that extend and impact the customer service of the logistics centre [146].
beyond the industrial sector and enable the development However, ensuring data privacy is just one of the obstacles
of new business models. With so much rivalry in today’s to overcome; lowering operational expenses is also crucial
market, businesses must constantly adapt to be competitive [147][148]RFID (radio-frequency identification. Many
in terms of cost, quality, and turnaround time [142]. Due questions have been raised about whether the application
to global competition, businesses need to be quick to adopt of these technologies in logistics centres would really cut or
new technologies and provide new items to the market [143]. increase costs. As an added argument, applying technologies
Designing productive, efficient, and adaptable techniques is in a uniform way, or prioritising each technology applied in
necessary to ensure process competitiveness throughout the logistics centres, would also be a topic for future research.
value chain [142]. The integration of new technologies into Therefore, in addition to the priority policies aimed at macro-
manufacturing processes, goods, and machinery is crucial development for countries, research directions for technology
to facilitating rapid response to changes in the marketplace application should also be explored in future studies of
[144]. One of the main obstacles faced by different countries logistics centres. Developing an acceptable application policy
is the lack of policies that allow technology to be applied to is a problem that requires further research.
the logistics sector. Fig. 8 illustrates the relationship between
logistics policy and governmental policies. It should be noted
that logistics policy does not have a superior attitude toward Conclusions and future prospects
other policies; rather, it implements some of these (such as
security and transportation policies) and cooperates with Exponential growth in the number of products that are
others (e.g. industrial and maritime policies). Nonetheless, it available has led to the persistent challenge of overload within
should be considered in each area of its functioning, including logistics centres and freight operations in the logistics sector.
in macroeconomic and detailed policies and (crucially) it Moreover, these activities not only result in the inefficient
should not be ‘closed’. This phenomenon is closely related use of time and resources but also have detrimental effects
to the processes of globalisation and aims to create a ‘no on the environment. Hence, the use of novel technology
barriers’ relationship between regions, states, and continents, arises as a viable approach to effectively address these issues
in terms of logistics [145]. in their entirety. The integration of advanced technologies,
such as Big Data, Blockchain, Internet of Things, and Cloud
Computing (commonly referred to as 4.0 technologies), has
been extensively studied and applied in various domains.
Through previous research and applications, this work
has demonstrated the significant potential of these 4.0
technologies in enhancing the operational capabilities and

efficiency of logistics centres, particularly in managing and Inf. Syst. Manag., vol. 29, no. 3, pp. 233–245, Jun. 2012,
controlling the movement of goods. doi: 10.1080/10580530.2012.687317.
The primary uses of each technology are as follows:
The Internet of Things (IoT) facilitates the collection of 4. Hao Wang, O. L. Osen, Guoyuan Li, Wei Li, Hong-Ning
a substantial volume of data by using RFID tags affixed to Dai, and Wei Zeng, “Big data and industrial Internet of
individual packages and sensors installed throughout the Things for the maritime industry in Northwestern Norway,”
whole of the logistics centre. The use of Big Data facilitates in TENCON 2015 - 2015 IEEE Region 10 Conference, Nov.
the processing of vast amounts of data in order to provide 2015, pp. 1–5. doi: 10.1109/TENCON.2015.7372918.
assistance for managerial decision-making. Blockchain
technology in the logistics centre provides transparency 5. V. D. Bui and H. P. Nguyen, “A Comprehensive Review on
across the whole of the workflow, including the arrival Big Data-Based Potential Applications in Marine Shipping
and departure of items. Cloud Computing facilitates the Management,” Int. J. Adv. Sci. Eng. Inf. Technol., vol. 11, no.
sharing of information while maintaining control, leading to 3, pp. 1067–1077, Jun. 2021, doi: 10.18517/ijaseit.11.3.15350.
continuous productivity improvements in logistics centres.
The use of these technologies has the potential to enhance the 6. S. Altendorfer-Kaiser, “The Influence of Big Data on
intelligence and advancement of a logistics centre. In addition, Production and Logistics,” 2017, pp. 221–227. doi:
it is recommended that future research is directed toward the 10.1007/978-3-319-66923-6_26.
development of a technologically advanced logistics centre
that conforms to established environmental criteria. Hence, 7. H. P. Nguyen, P. Q. P. Nguyen, and V. D. Bui, “Applications
it can be inferred that the adoption of a green-smart logistics of Big Data Analytics in Traffic Management in Intelligent
centre model will become an unavoidable trajectory in the Transportation Systems,” JOIV Int. J. Informatics Vis.,
future. This model would enable the systematic regulation vol. 6, no. 1–2, pp. 177–187, May 2022, doi: 10.30630/
of goods transportation to effectively cater to the substantial joiv.6.1-2.882.
demand for commodities. Furthermore, these logistics
hubs have the potential to function as seamless extensions 8. C. S. Tang and L. P. Veelenturf, “The strategic role of
of seaports, therefore mitigating congestion issues in port logistics in the industry 4.0 era,” Transp. Res. Part E Logist.
cities. Transportation operations within the logistics business Transp. Rev., vol. 129, pp. 1–11, Sep. 2019, doi: 10.1016/j.
are well acknowledged as being a significant contributor to tre.2019.06.004.
environmental degradation. The escalation of congestion
and prolonged waiting times at conventional logistics 9. A. Ferrari, G. Mangano, A. C. Cagliano, and A. De
hubs and seaports has led to a significant rise in pollution Marco, “4.0 Technologies in City Logistics: an Empirical
levels, hence posing a substantial urban challenge. The use Investigation of Contextual Factors,” Oper. Manag.
of 4.0 technologies enables the expeditious regulation of Res., vol. 16, no. 1, pp. 345–362, 2023, doi: 10.1007/
products inside centres, resulting in reduced waiting times s12063-022-00304-5.
and, subsequently, contributing to the mitigation of vehicle
emissions. Hence, the establishment of a green and intelligent 10. H. Golpîra, S. A. R. Khan, and S. Safaeipour, “A review
logistics centre emerges as a prominent research direction of logistics Internet-of-Things: Current trends and scope
aimed at mitigating environmental pollution. Simultaneously, for future research,” J. Ind. Inf. Integr., vol. 22, p. 100194,
this development aligns with the evolving requirements of Jun. 2021, doi: 10.1016/j.jii.2020.100194.
smart cities, which are experiencing rapid global growth.
11. D. Peraković, M. Periša, and P. Zorić, “Challenges and
Issues of ICT in Industry 4.0,” 2020, pp. 259–269. doi:
References 10.1007/978-3-030-22365-6_26.
1. L. Atzori, A. Iera, and G. Morabito, “The internet of 12. K. Witkowski, “Internet of Things, Big Data, Industry
things: A survey,” Comput. networks, vol. 54, no. 15, pp. 4.0 – Innovative Solutions in Logistics and Supply Chains
2787–2805, 2010. Management,” Procedia Eng., vol. 182, pp. 763–769, 2017,
doi: 10.1016/j.proeng.2017.03.197.
2. D. Bandyopadhyay and J. Sen, “Internet of Things:
Applications and Challenges in Technology and 13. H. P. Nguyen, P. Q. P. Nguyen, D. K. P. Nguyen, V. D.
Standardization,” Wirel. Pers. Commun., vol. 58, no. 1, Bui, and D. T. Nguyen, “Application of IoT Technologies
pp. 49–69, May 2011, doi: 10.1007/s11277-011-0288-5. in Seaport Management,” JOIV Int. J. Informatics Vis.,
vol. 7, no. 1, p. 228, Mar. 2023, doi: 10.30630/joiv.7.1.1697.
3. C. Bardaki, P. Kourouthanassis, and K. Pramatari,
“Deploying RFID-Enabled Services in the Retail Supply 14. A. Bekrar, A. Ait El Cadi, R. Todosijevic, and J. Sarkis,
Chain: Lessons Learned toward the Internet of Things,” “Digitalizing the Closing-of-the-Loop for Supply
Chains: A Transportation and Blockchain Perspective,”

Sustainability, vol. 13, no. 5, p. 2895, Mar. 2021, doi: of Things: A Dueling Deep ${Q}$ -Learning Approach,”
10.3390/su13052895. IEEE Internet Things J., vol. 6, no. 3, pp. 4627–4639, Jun.
2019, doi: 10.1109/JIOT.2018.2871394.
15. M. Kouhizadeh, S. Saberi, and J. Sarkis, “Blockchain
technology and the sustainable supply chain: Theoretically 26. K. Ashton, “That ‘internet of things’ thing,” RFID J., vol.
exploring adoption barriers,” Int. J. Prod. Econ., vol. 231, 22, no. 7, pp. 97–114, 2009.
p. 107831, Jan. 2021, doi: 10.1016/j.ijpe.2020.107831.
27. K. Witkowski, “Internet of Things, Big Data, Industry
16. A. A. Mukherjee, R. K. Singh, R. Mishra, and S. Bag, 4.0 - Innovative Solutions in Logistics and Supply Chains
“Application of blockchain technology for sustainability Management,” Procedia Eng., vol. 182, pp. 763–769, 2017,
development in agricultural supply chain: justification doi: 10.1016/j.proeng.2017.03.197.
framework,” Oper. Manag. Res., vol. 15, no. 1–2, pp. 46–61,
Jun. 2022, doi: 10.1007/s12063-021-00180-5. 28. “Discussion: Will Warehouse Robots Completely Replace
Traditional Logistics Industry Model?,” Geekplus, 2018.
17. M. A. Ahad, S. Paiva, G. Tripathi, and N. Feroz, “Enabling
technologies and sustainable smart cities,” Sustain. 29. R. Bergqvist, “Transport and logistics facilities expansion
Cities Soc., vol. 61, p. 102301, Oct. 2020, doi: 10.1016/j. and social sustainability: A critical discussion and findings
scs.2020.102301. from the City of Gothenburg, Swe-den.” Gothenburg:
Tesis Doctoral, 2016.
18. S. Bhardwaj, L. Jain, and S. Jain, “Cloud computing:
A study of infrastructure as a service (IAAS),” Int. J. Eng. 30. J. T. Bowen, “Moving places: the geography of warehousing
Inf. Technol., vol. 2, no. 1, pp. 60–63, 2010. in the US,” J. Transp. Geogr., vol. 16, no. 6, pp. 379–387,
Nov. 2008, doi: 10.1016/j.jtrangeo.2008.03.001.
19. T.-M. Choi, S. W. Wallace, and Y. Wang, “Big Data
Analytics in Operations Management,” Prod. Oper. 31. A. McKinnon, “The present and future land requirements
Manag., vol. 27, no. 10, pp. 1868–1883, Oct. 2018, doi: of logistical activities,” Land use policy, vol. 26, pp. S293–
10.1111/poms.12838. S301, Dec. 2009, doi: 10.1016/j.landusepol.2009.08.014.
20. D. Zhu, “IOT and big data based cooperative logistical 32. Q. Yuan, “Environmental Justice in Warehousing
delivery scheduling method and cloud robot system,” Location,” J. Plan. Lit., vol. 33, no. 3, pp. 287–298, Aug.
Futur. Gener. Comput. Syst., vol. 86, pp. 709–715, Sep. 2018, doi: 10.1177/0885412217753841.
2018, doi: 10.1016/j.future.2018.04.081.
33. Q. Yuan, “Location of Warehouses and Environmental
21. T. T. Le et al., “Management strategy for seaports aspiring Justice,” J. Plan. Educ. Res., vol. 41, no. 3, pp. 282–293,
to green logistical goals of IMO: Technology and policy Sep. 2021, doi: 10.1177/0739456X18786392.
solutions,” Polish Marit. Res., vol. 30, no. 2, pp. 165–187,
2023, doi: 10.2478/pomr-2023-0031. 34. Q. Yuan and J. Zhu, “Logistics sprawl in Chinese
metropolises: Evidence from Wuhan,” J. Transp.
22. E. J. Khatib and R. Barco, “Optimization of 5G Networks Geogr., vol. 74, pp. 242–252, Jan. 2019, doi: 10.1016/j.
for Smart Logistics,” Energies, vol. 14, no. 6, p. 1758, Mar. jtrangeo.2018.11.019.
2021, doi: 10.3390/en14061758.
35. F. Straube, A. L. Junge, P. Verhoeven, J. Reipert, and M.
23. J.-Q. Li, F. R. Yu, G. Deng, C. Luo, Z. Ming, and Q. Mansfeld, Pathway of digital transformation in logistics :
Yan, “Industrial Internet: A Survey on the Enabling best practice concepts and future developments, no. July.
Technologies, Applications, and Challenges,” IEEE 2019. doi: 10.14279/depositonce-8502.
Commun. Surv. Tutorials, vol. 19, no. 3, pp. 1504–1526,
2017, doi: 10.1109/COMST.2017.2691349. 36. R. Narasimhan and S. W. Kim, “INFORMATION
SYSTEM UTILIZATION STRATEGY FOR SUPPLY
24. M. Liu, F. R. Yu, Y. Teng, V. C. M. Leung, and M. Song, CHAIN INTEGRATION,” J. Bus. Logist., vol. 22, no. 2, pp.
“Performance Optimization for Blockchain-Enabled 51–75, Sep. 2001, doi: 10.1002/j.2158-1592.2001.tb00003.x.
Industrial Internet of Things (IIoT) Systems: A Deep
Reinforcement Learning Approach,” IEEE Trans. Ind. 37. S. Kot, I. Goldbach, and B. Ślusarczyk, “Supply chain
Informatics, vol. 15, no. 6, pp. 3559–3570, Jun. 2019, doi: management in SMEs – Polish and Romanian approach,”
10.1109/TII.2019.2897805. Econ. Sociol., vol. 11, no. 4, pp. 142–156, Dec. 2018, doi:
10.14254/2071-789X.2018/11-4/9.
25. C. Qiu, F. R. Yu, H. Yao, C. Jiang, F. Xu, and C. Zhao,
“Blockchain-Based Software-Defined Industrial Internet

38. M. Levinson, The box: how the shipping container made 49. A. T. Hoang et al., “Energy-related approach for reduction
the world smaller and the world economy bigger. Princeton of CO2 emissions: A critical strategy on the port-to-ship
University Press, 2016. pathway,” J. Clean. Prod., vol. 355, p. 131772, Jun. 2022,
doi: 10.1016/j.jclepro.2022.131772.
39. V. D. Bui and H. P. Nguyen, “The role of the inland
container depot system in developing a sustainable 50. S. Vakili, A. I. Ölçer, A. Schönborn, F. Ballini, and A.
transport system,” Int. J. Knowledge-Based Dev., vol. 12, no. T. Hoang, “Energy‐related clean and green framework
3/4, pp. 424–443, 2022, doi: 10.1504/IJKBD.2022.128914. for shipbuilding community towards zero‐emissions:
A strategic analysis from concept to case study,” Int. J.
40. V. Aulin, A. Hrynkiv, S. Lysenko, A. Dykha, T. Zamota, Energy Res., vol. 46, no. 14, pp. 20624–20649, Nov. 2022,
and V. Dzyura, “Exploring a possibility to control the doi: 10.1002/er.7649.
stressedstrained state of cylinder liners in diesel engines
by the tribotechnology of alignment,” Eastern-European 51. C. Altuntaş and O. Tuna, “Greening Logistics Centers: The
J. Enterp. Technol., vol. 3, no. 12 (99), pp. 6–16, Jun. 2019, Evolution of Industrial Buying Criteria Towards Green,”
doi: 10.15587/1729-4061.2019.171619. Asian J. Shipp. Logist., vol. 29, no. 1, pp. 59–80, Apr. 2013,
doi: 10.1016/j.ajsl.2013.05.004.
41. V. Aulin, A. Hrinkiv, A. Dykha, M. Chernovol, O.
Lyashuk, and S. Lysenko, “Substantiation of diagnostic 52. C. Uyanık, G. Tuzkaya, S. Oguztimur, C. Uyanik, and S.
parameters for determining the technical condition of Oğuzti̇mur, “A Literature Survey On Logistics Centers’
transmission assemblies in trucks,” Eastern-European J. Location Selection Problem,” Sigma J. Eng. Nat. Sci., vol.
Enterp. Technol., vol. 2, no. 1 (92), pp. 4–13, Mar. 2018, 36, no. 1, pp. 141–160, 2018.
doi: 10.15587/1729-4061.2018.125349.
53. B. Koldemir, M. Çanci, and E. Gönüler, “Büyük ölçekli kent
42. J.-P. Rodrigue, The Geography of Transport Systems. Fifth planlamasında lojistik köyler,” İzmir Ulaşım Sempozyumu,
edition. | Abingdon, Oxon ; New York, NY : Routledge, pp. 8–9, 2009.
2020.: Routledge, 2020. doi: 10.4324/9780429346323.
54. C. D. Higgins and M. R. Ferguson, “An Exploration of the
43. F. Weidinger, N. Boysen, and M. Schneider, “Picker routing Freight Village Concept and its Applicability to Ontario,”
in the mixed-shelves warehouses of e-commerce retailers,” 2011.
Eur. J. Oper. Res., vol. 274, no. 2, pp. 501–515, Apr. 2019,
doi: 10.1016/j.ejor.2018.10.021. 55. K. Rimienė and D. Grundey, “Logistics centre concept
through evolution and definition,” Eng. Econ., vol. 54, no.
44. V. Aulin et al., “Increasing the Functioning Efficiency of 4, pp. 87–95, 2007.
the Working Warehouse of the ‘UVK Ukraine’ Company
Transport and Logistics Center,” Commun. - Sci. Lett. Univ. 56. B. Erkayman, E. Gundogar, G. Akkaya, and M. Ipek,
Zilina, vol. 22, no. 2, pp. 3–14, Apr. 2020, doi: 10.26552/ “A Fuzzy Topsis Approach For Logistics Center Location
com.C.2020.2.3-14. Selection,” J. Bus. Case Stud., vol. 7, no. 3, pp. 49–54, Apr.
2011, doi: 10.19030/jbcs.v7i3.4263.
45. F. H. Staudt, G. Alpan, M. Di Mascolo, and C. M. T.
Rodriguez, “Warehouse performance measurement: 57. H. P. Nguyen, P. Q. P. Nguyen, and T. P. Nguyen, “Green
a literature review,” Int. J. Prod. Res., vol. 53, no. 18, pp. Port Strategies in Developed Coastal Countries as Useful
5524–5544, Sep. 2015, doi: 10.1080/00207543.2015.1030466. Lessons for the Path of Sustainable Development: A case
study in Vietnam,” Int. J. Renew. Energy Dev., vol. 11, no.
46. A. Nettsträter, T. Geißen, M. Witthaut, D. Ebel, and J. 4, pp. 950–962, Nov. 2022, doi: 10.14710/ijred.2022.46539.
Schoneboom, “Logistics Software Systems and Functions:
An Overview of ERP, WMS, TMS and SCM Systems,” 2015, 58. İ. ÖNDEN, F. Eldemir, and M. Canci, “Logistics center
pp. 1–11. doi: 10.1007/978-3-319-13404-8_1. concept and location decision criteria,” Sigma J. Eng. Nat.
Sci., vol. 33, no. 3, pp. 325–340, 2015.
47. “Logistics Centers Directions For Use,” EUROPLATFORMS
EEIG, 2004. 59. R. Kaynak, İ. Koçoğlu, and A. E. Akgün, “The Role of
Reverse Logistics in the Concept of Logistics Centers,”
48. A. T. Hoang et al., “Technological solutions for boosting Procedia - Soc. Behav. Sci., vol. 109, pp. 438–442, Jan. 2014,
hydrogen role in decarbonization strategies and net-zero doi: 10.1016/j.sbspro.2013.12.487.
goals of world shipping: Challenges and perspectives,”
Renew. Sustain. Energy Rev., vol. 188, p. 113790, Dec. 2023, 60. H. P. Nguyen, “Blockchain-an indispensable development
doi: 10.1016/j.rser.2023.113790. trend of logistics industry in Vietnam: Current situation

and recommended solutions,” Int. J. e-Navigation Marit. warehouse operations,” Expert Syst. Appl., vol. 30, no. 4,
Econ., vol. 13, pp. 14–22, 2019. pp. 561–576, May 2006, doi: 10.1016/j.eswa.2005.07.023.
61. V. D. Bui and H. P. Nguyen, “A Systematized Review on 73. V. Sharma, I. You, G. Pau, M. Collotta, J. Lim, and J.
Rationale and Experience to Develop Advanced Logistics Kim, “LoRaWAN-Based Energy-Efficient Surveillance by
Center System in Vietnam,” Webology, vol. 18, pp. 89–101, Drones for Intelligent Transportation Systems,” Energies,
2021. vol. 11, no. 3, p. 573, Mar. 2018, doi: 10.3390/en11030573.
62. H. Makino, K. Tamada, K. Sakai, and S. Kamijo, “Solutions 74. R. H. Weber and R. Weber, “Governance of the Internet
for urban traffic issues by ITS technologies,” IATSS of Things,” in Internet of Things, Berlin, Heidelberg:
Res., vol. 42, no. 2, pp. 49–60, Jul. 2018, doi: 10.1016/j. Springer Berlin Heidelberg, 2010, pp. 69–100. doi:
iatssr.2018.05.003. 10.1007/978-3-642-11710-7_4.
63. S. Lee, Y. Kang, and V. V. Prabhu, “Smart logistics: 75. X. Feng, L. T. Yang, L. Wang, and A. Vinel, “Internet of
distributed control of green crowdsourced parcel services,” things,” Int. J. Commun. Syst., vol. 25, no. 9, pp. 1101–1102,
Int. J. Prod. Res., vol. 54, no. 23, pp. 6956–6968, Dec. 2016, 2012.
doi: 10.1080/00207543.2015.1132856.
76. Robert Sheldon, “Definition Sensor,” 2022.
64. A. Kawa, “SMART Logistics Chain,” in Intelligent
Information and Database Systems, 2012, pp. 432–438. 77. M. Wang, C. C. Wang, S. Sepasgozar, and S. Zlatanova,
doi: 10.1007/978-3-642-28487-8_45. “A Systematic Review of Digital Technology Adoption
in Off-Site Construction: Current Status and Future
65. H. P. Nguyen, “Sustainable development of logistics in Direction towards Industry 4.0,” Buildings, vol. 10, no.
Vietnam in the period 2020-2025,” Int. J. Innov. Creat. 11, p. 204, Nov. 2020, doi: 10.3390/buildings10110204.
Chang., vol. 11, no. 3, pp. 665–682, 2020.
78. C. Mao, X. Tao, H. Yang, R. Chen, and G. Liu, “Real-
66. J. Mehmann and F. Teuteberg, “Process reengineering by time carbon emissions monitoring tool for prefabricated
using the 4PL approach,” Bus. Process Manag. J., vol. 22, no. construction: An iot-based system framework,” in
4, pp. 879–902, Jul. 2016, doi: 10.1108/BPMJ-12-2014-0119. International Conference on Construction and Real Estate
Management 2018, 2018, pp. 121–127.
67. Y. M. Tang, G. T. S. Ho, Y. Y. Lau, and S. Y. Tsui, “Integrated
Smart Warehouse and Manufacturing Management with 79. G. Liu et al., “Cyber-physical system-based real-time
Demand Forecasting in Small-Scale Cyclical Industries,” monitoring and visualization of greenhouse gas emissions
Machines, vol. 10, no. 6, 2022. of prefabricated construction,” J. Clean. Prod., vol. 246,
p. 119059, Feb. 2020, doi: 10.1016/j.jclepro.2019.119059.
68. V. Yavas and Y. D. Ozkan-Ozen, “Logistics centers in the
new industrial era: A proposed framework for logistics 80. Y. Zhai et al., “An Internet of Things-enabled BIM platform
center 4.0,” Transp. Res. Part E Logist. Transp. Rev., vol. for modular integrated construction: A case study in Hong
135, p. 101864, Mar. 2020, doi: 10.1016/j.tre.2020.101864. Kong,” Adv. Eng. Informatics, vol. 42, p. 100997, Oct. 2019,
doi: 10.1016/j.aei.2019.100997.
69. S. Winkelhaus and E. H. Grosse, “Logistics 4.0:
a systematic review towards a new logistics system,” 81. M. Valinejadshoubi, A. Bagchi, and O. Moselhi,
Int. J. Prod. Res., vol. 58, no. 1, pp. 18–43, Jan. 2020, doi: “Development of a BIM-Based Data Management System
10.1080/00207543.2019.1612964. for Structural Health Monitoring with Application
to Modular Buildings: Case Study,” J. Comput. Civ.
70. G. S. Cho, “A Study on Establishment of Smart Logistics Eng., vol. 33, no. 3, May 2019, doi: 10.1061/(ASCE)
Center based on Logistics 4.0,” J. Multimed. Inf. Syst., vol. CP.1943-5487.0000826.
5, no. 4, pp. 265-272., 2018, doi: http://dx.doi.org/10.9717/
JMIS.2018.5.4.265. 82. my Murata, “Barometric Pressure Sensor Basics.”
71. W. Hamdy, N. Mostafa, and H. Elawady, “Towards a smart 83. X. Li, H. Chi, P. Wu, and G. Q. Shen, “Smart work
warehouse management system,” Proc. Int. Conf. Ind. Eng. packaging-enabled constraint-free path re-planning for
Oper. Manag., vol. 2018, no. SEP, pp. 2555–2563, 2018. tower crane in prefabricated products assembly process,”
Adv. Eng. Informatics, vol. 43, p. 101008, Jan. 2020, doi:
72. H. CHOW, K. CHOY, W. LEE, and K. LAU, “Design 10.1016/j.aei.2019.101008.
of a RFID case-based resource management system for

84. Z. Wang, H. Hu, and W. Zhou, “RFID Enabled Knowledge- 95. G.-S. Cho, “A study on establishment of smart logistics
Based Precast Construction Supply Chain,” Comput. Civ. center based on logistics 4.0,” J. Multimed. Inf. Syst., vol.
Infrastruct. Eng., vol. 32, no. 6, pp. 499–514, Jun. 2017, doi: 5, no. 4, pp. 265–272, 2018.
10.1111/mice.12254.
96. W. Wu, C. Cheung, S. Y. Lo, R. Y. Zhong, and G. Q. Huang,
85. D. Uckelmann, M. Harrison, and F. Michahelles, “An “An IoT-enabled Real-time Logistics System for A Third
Architectural Approach Towards the Future Internet Party Company: A Case Study,” Procedia Manuf., vol. 49,
of Things,” in Architecting the Internet of Things, Berlin, pp. 16–23, 2020, doi: 10.1016/j.promfg.2020.06.005.
Heidelberg: Springer Berlin Heidelberg, 2011, pp. 1–24.
doi: 10.1007/978-3-642-19157-2_1. 97. Oleksii Kholodenko, “IOT IN WAR EHOUSE
MANAGEMENT — EXTENSIVE GUIDE,” 2022.
86. R. K. Chahal, N. Kumar, and S. Batra, “Trust management
in social Internet of Things: A taxonomy, open issues, and 98. C. K. M. Lee, Y. Lv, K. K. H. Ng, W. Ho, and K. L. Choy,
challenges,” Comput. Commun., vol. 150, pp. 13–46, Jan. “Design and application of Internet of things-based
2020, doi: 10.1016/j.comcom.2019.10.034. warehouse management system for smart logistics,” Int.
J. Prod. Res., vol. 56, no. 8, pp. 2753–2768, Apr. 2018, doi:
87. D. Miorandi, S. Sicari, F. De Pellegrini, and I. Chlamtac, 10.1080/00207543.2017.1394592.
“Internet of things: Vision, applications and research
challenges,” Ad Hoc Networks, vol. 10, no. 7, pp. 1497–1516, 99. W. Hamdy, N. Mostafa, and H. Elawady, “Towards a smart
Sep. 2012, doi: 10.1016/j.adhoc.2012.02.016. warehouse management system,” in Proceedings of the
International Conference on Industrial Engineering and
88. D. Mishra, A. Gunasekaran, S. J. Childe, T. Papadopoulos, Operations Management, 2018, pp. 2555–2563.
R. Dubey, and S. Wamba, “Vision, applications and future
challenges of Internet of Things,” Ind. Manag. Data Syst., 100. M. Oner, A. Budak, and A. Ustundag, “RFID-based
vol. 116, no. 7, pp. 1331–1355, Aug. 2016, doi: 10.1108/ warehouse management system in wool yarn industry,”
IMDS-11-2015-0478. Int. J. RF Technol., vol. 8, no. 4, pp. 165–189, Feb. 2018,
doi: 10.3233/RFT-171655.
89. I. C. L. Ng and S. Y. L. Wakenshaw, “The Internet-of-
Things: Review and research directions,” Int. J. Res. 101. B. S. S. Tejesh and S. Neeraja, “Warehouse inventory
Mark., vol. 34, no. 1, pp. 3–21, Mar. 2017, doi: 10.1016/j. management system using IoT and open source
ijresmar.2016.11.003. framework,” Alexandria Eng. J., vol. 57, no. 4, pp. 3817–
3823, Dec. 2018, doi: 10.1016/j.aej.2018.02.003.
90. A. Whitmore, A. Agarwal, and L. Da Xu, “The Internet of
Things—A survey of topics and trends,” Inf. Syst. Front., 102. C.-C. Lin and J.-W. Yang, “Cost-Efficient Deployment of
2015, doi: 10.1007/s10796-014-9489-2. Fog Computing Systems at Logistics Centers in Industry
4.0,” IEEE Trans. Ind. Informatics, vol. 14, no. 10, pp. 4603–
91. L. Da Xu, W. He, and S. Li, “Internet of Things in Industries: 4611, Oct. 2018, doi: 10.1109/TII.2018.2827920.
A Survey,” IEEE Trans. Ind. Informatics, vol. 10, no. 4, pp.
2233–2243, Nov. 2014, doi: 10.1109/TII.2014.2300753. 103. B. Feng and Q. Ye, “Operations management of smart
logistics: A literature review and future research,” Front.
92. Y. Ding, M. Jin, S. Li, and D. Feng, “Smart logistics based Eng. Manag., vol. 8, no. 3, pp. 344–355, Sep. 2021, doi:
on the internet of things technology: an overview,” Int. 10.1007/s42524-021-0156-2.
J. Logist. Res. Appl., vol. 24, no. 4, pp. 323–345, Jul. 2021,
doi: 10.1080/13675567.2020.1757053. 104. Y. Li, F. Chu, C. Feng, C. Chu, and M. Zhou, “Integrated
Production Inventory Routing Planning for Intelligent
93. R. Y. Zhong, Q. Y. Dai, T. Qu, G. J. Hu, and G. Q. Huang, Food Logistics Systems,” IEEE Trans. Intell. Transp.
“RFID-enabled real-time manufacturing execution system Syst., vol. 20, no. 3, pp. 867–878, Mar. 2019, doi: 10.1109/
for mass-customization production,” Robot. Comput. TITS.2018.2835145.
Integr. Manuf., vol. 29, no. 2, pp. 283–292, Apr. 2013, doi:
10.1016/j.rcim.2012.08.001. 105. N. Zhang, “Smart Logistics Path for Cyber-Physical
Systems With Internet of Things,” IEEE Access, vol. 6, pp.
94. X. T. R. Kong, X. Yang, G. Q. Huang, and H. Luo, “The 70808–70819, 2018, doi: 10.1109/ACCESS.2018.2879966.
impact of industrial wearable system on industry 4.0,” in
2018 IEEE 15th International Conference on Networking, 106. S. Jagtap, F. Bader, G. Garcia-Garcia, H. Trollman, T.
Sensing and Control (ICNSC), Mar. 2018, pp. 1–6. doi: Fadiji, and K. Salonitis, “Food Logistics 4.0: Opportunities
10.1109/ICNSC.2018.8361266. and Challenges,” Logistics, vol. 5, no. 1, p. 2, Dec. 2020,
doi: 10.3390/logistics5010002.

107. Q. Zhang, L. Shi, and S. Sun, “Optimization of Intelligent 118. G. T. S. Ho, Y. M. Tang, K. Y. Tsang, V. Tang, and K.
Logistics System Based on Big Data Collection Techniques,” Y. Chau, “A blockchain-based system to enhance
2023, pp. 378–387. doi: 10.1007/978-3-031-31860-3_40. aircraft parts traceability and trackability for inventory
management,” Expert Syst. Appl., vol. 179, p. 115101, Oct.
108. M. E. Xie and Q. Ou, “Explore the Application of Big Data 2021, doi: 10.1016/j.eswa.2021.115101.
Technology in Modern Enterprise Logistics Management,”
2023, pp. 150–161. doi: 10.1007/978-3-031-24468-1_14. 119. B. Al Kurdi, H. M. Alzoubi, I. Akour, and M. T. Alshurideh,
“The effect of blockchain and smart inventory system on
109. G. Wang, A. Gunasekaran, and E. W. T. Ngai, “Distribution supply chain performance: Empirical evidence from retail
network design with big data: model and analysis,” Ann. industry,” Uncertain Supply Chain Manag., vol. 10, no. 4,
Oper. Res., vol. 270, no. 1–2, pp. 539–551, Nov. 2018, doi: pp. 1111–1116, 2022, doi: 10.5267/j.uscm.2022.9.001.
10.1007/s10479-016-2263-8.
120. G. V. Lakshmi, S. Gogulamudi, B. Nagaeswari, and S.
110. G. Wang, A. Gunasekaran, E. W. T. Ngai, and T. Reehana, “BlockChain Based Inventory Management by
Papadopoulos, “Big data analytics in logistics and supply QR Code Using Open CV,” in 2021 International Conference
chain management: Certain investigations for research on Computer Communication and Informatics (ICCCI),
and applications,” Int. J. Prod. Econ., vol. 176, pp. 98–110, Jan. 2021, pp. 1–6. doi: 10.1109/ICCCI50826.2021.9402666.
Jun. 2016, doi: 10.1016/j.ijpe.2016.03.014.
121. T. Mladenović, “Cloud Computing in logistics,” University
111. A. A. C. Vieira, L. Pedro, M. Y. Santos, J. M. Fernandes, of Belgrade, 2018.
and L. S. Dias, “Data Requirements Elicitation in
Big Data Warehousing,” 2019, pp. 106–113. doi: 122.G. R and L. M, “The Concept of Logistics 4.0,” 4th Logist.
10.1007/978-3-030-11395-7_10. Int. Conf., pp. 283–293, 2019.
112. M. Zheng and K. Wu, “Smart spare parts management 123. Accenture, “The state of of supply The state supply chain
systems in semiconductor manufacturing,” Ind. Manag. and the cloud chain and the cloud,” 2022.
Data Syst., vol. 117, no. 4, pp. 754–763, May 2017, doi:
10.1108/IMDS-06-2016-0242. 124.Maersk, “How cloud computing is shaping the future of
logistics,” 2023.
113. S. N. Wahab, M. I. Hamzah, N. M. Sayuti, W. C. Lee,
and S. Y. Tan, “Big data analytics adoption: an empirical 125. Y. Jiang, “Location and Path Planning of Cross-Border
study in the Malaysian warehousing sector,” Int. J. Logist. E-Commerce Logistics Distribution Center in Cloud
Syst. Manag., vol. 40, no. 1, p. 121, 2021, doi: 10.1504/ Computing Environment,” 2021, pp. 30–40. doi:
IJLSM.2021.117703. 10.1007/978-3-030-67871-5_4.
114. R. W. Ahmad, H. Hasan, R. Jayaraman, K. Salah, and 126.X. Zhang, “Optimization design of railway logistics
M. Omar, “Blockchain applications and architectures for center layout based on mobile cloud edge computing,”
port operations and logistics management,” Res. Transp. PeerJ Comput. Sci., vol. 9, p. e1298, Apr. 2023, doi: 10.7717/
Bus. Manag., vol. 41, p. 100620, Dec. 2021, doi: 10.1016/j. peerj-cs.1298.
rtbm.2021.100620.
127. P. Sharma and S. Panda, “Cloud Computing for Supply
115. S. Johar, N. Ahmad, W. Asher, H. Cruickshank, and A. Chain Management and Warehouse Automation: A Case
Durrani, “Research and Applied Perspective to Blockchain Study of Azure Cloud,” Int. J. Smart Sens. Adhoc Network.,
Technology: A Comprehensive Survey,” Appl. Sci., vol. 11, pp. 19–29, Jan. 2023, doi: 10.47893/IJSSAN.2023.1227.
no. 14, p. 6252, Jul. 2021, doi: 10.3390/app11146252.
128.B. Gupta, A. Gupta, and A. Nagpal, “Implementation of
116. V. Aleksieva, H. Valchanov, A. Haka, and D. Dinev, Warehouse Management Through Cloud Computing,”
“Logistics Model Based on Smart Contracts on Blockchain IJIRST-International J. Innov. Res. Sci. Technol., vol. 3, no.
and IoT,” in EEPES’23, Jul. 2023, p. 8. doi: 10.3390/ 10, pp. 189–192, 2017.
engproc2023041008.
129. V. Sivakumar, R. Ruthramathi, and S. Leelapriyadharsini,
117. J. Chen, S. Xu, K. Liu, S. Yao, X. Luo, and H. Wu, “Intelligent “Challenges of Cloud Computing in Warehousing
Transportation Logistics Optimal Warehouse Location Operations with Respect to Chennai Port Trust,” in
Method Based on Internet of Things and Blockchain Proceedings of the 2020 the 3rd International Conference
Technology,” Sensors, vol. 22, no. 4, p. 1544, Feb. 2022, on Computers in Management and Business, Jan. 2020,
doi: 10.3390/s22041544. pp. 162–165. doi: 10.1145/3383845.3383895.

130. L. Barreto, A. Amaral, and T. Pereira, “Industry 4.0 Procedia Manuf., vol. 11, pp. 796–805, 2017, doi: 10.1016/j.
implications in logistics: an overview,” Procedia Manuf., vol. promfg.2017.07.181.
13, pp. 1245–1252, 2017, doi: 10.1016/j.promfg.2017.09.045.
142. U. Dombrowski, T. Richter, and P. Krenkel,
131. M. Vanderroost, P. Ragaert, J. Verwaeren, B. De Meulenaer, “Interdependencies of Industrie 4.0 & Lean Production
B. De Baets, and F. Devlieghere, “The digitization of a food Systems: A Use Cases Analysis,” Procedia Manuf., vol. 11,
package’s life cycle: Existing and emerging computer pp. 1061–1068, 2017, doi: 10.1016/j.promfg.2017.07.217.
systems in the pre-logistics phase,” Comput. Ind., vol. 87,
pp. 1–14, May 2017, doi: 10.1016/j.compind.2017.02.002. 143. F. Hecklau, M. Galeitzke, S. Flachs, and H. Kohl, “Holistic
Approach for Human Resource Management in Industry
132. J. Yan et al., “Intelligent Supply Chain Integration and 4.0,” Procedia CIRP, vol. 54, pp. 1–6, 2016, doi: 10.1016/j.
Management Based on Cloud of Things,” Int. J. Distrib. procir.2016.05.102.
Sens. Networks, vol. 10, no. 3, p. 624839, Mar. 2014, doi:
10.1155/2014/624839. 144. A. G. Pereira*, T. M. Lima, and F. Charrua-Santos,
“Industry 4.0 and Society 5.0: Opportunities and Threats,”
133. Y. P. Tsang, K. L. Choy, C.-H. Wu, G. T. S. Ho, H. Y. Lam, Int. J. Recent Technol. Eng., vol. 8, no. 5, pp. 3305–3308,
and V. Tang, “An intelligent model for assuring food Jan. 2020, doi: 10.35940/ijrte.D8764.018520.
quality in managing a multi-temperature food distribution
centre,” Food Control, vol. 90, pp. 81–97, Aug. 2018, doi: 145. J. Korczak and K. Kijewska, “Smart Logistics in the
10.1016/j.foodcont.2018.02.030. development of Smart Cities,” Transp. Res. Procedia, vol.
39, pp. 201–211, 2019, doi: 10.1016/j.trpro.2019.06.022.
134.P. J. B. Sanchez et al., “Use of UIoT for Offshore Surveys
Through Autonomous Vehicles,” Polish Marit. Res., vol. 28, 146. D. S. Terzi, R. Terzi, and S. Sagiroglu, “A survey on
no. 3, pp. 175–189, Sep. 2021, doi: 10.2478/pomr-2021-0044. security and privacy issues in big data,” in 2015 10th
International Conference for Internet Technology and
135. M. Drzewiecki and J. Guziński, “Design of an Autonomous Secured Transactions (ICITST), Dec. 2015, pp. 202–207.
IoT Node Powered by a Perovskite-Based Wave Energy doi: 10.1109/ICITST.2015.7412089.
Converter,” Polish Marit. Res., vol. 30, no. 3, pp. 142–152,
Sep. 2023, doi: 10.2478/pomr-2023-0047. 147. V. N. Inukollu, S. Arsi, and S. Rao Ravuri, “Security Issues
Associated with Big Data in Cloud Computing,” Int. J.
136. N. D. K. Pham, G. H. Dinh, H. T. Pham, J. Kozak, and H. Netw. Secur. Its Appl., vol. 6, no. 3, pp. 45–56, May 2014,
P. Nguyen, “Role of Green Logistics in the Construction of doi: 10.5121/ijnsa.2014.6304.
Sustainable Supply Chains,” Polish Marit. Res., vol. 30, no.
3, pp. 191–211, Sep. 2023, doi: 10.2478/pomr-2023-0052. 148. K. Buntak, M. Kovačić, and M. Mutavdžija, “Internet of
things and smart warehouses as the future of logistics,”
137. A. Oke and M. Gopalakrishnan, “Managing disruptions Teh. Glas., vol. 13, no. 3, pp. 248–253, Sep. 2019, doi:
in supply chains: A case study of a retail supply chain,” 10.31803/tg-20190215200430.
Int. J. Prod. Econ., vol. 118, no. 1, pp. 168–174, Mar. 2009,
doi: 10.1016/j.ijpe.2008.08.045.
138. T. K. Dasaklis, C. P. Pappis, and N. P. Rachaniotis,

“Epidemics control and logistics operations: A review,”
Int. J. Prod. Econ., vol. 139, no. 2, pp. 393–410, Oct. 2012,
doi: 10.1016/j.ijpe.2012.05.023.
139. S. Columbus, “Honesty-Humility, Beliefs, and Prosocial

Behaviour: A Test on Stockpiling During the COVID-19
Pandemic,” Collabra Psychol., vol. 7, no. 1, Feb. 2021, doi:
10.1525/collabra.19028.
140. A. C. Pereira and F. Romero, “A review of the meanings

and the implications of the Industry 4.0 concept,” Procedia
Manuf., vol. 13, pp. 1206–1214, 2017, doi: 10.1016/j.
promfg.2017.09.032.
141. H. Unger, F. Börner, and E. Müller, “Context Related

Information Provision in Industry 4.0 Environments,”

PMRes 2023 4

Uploaded by

Copyright:

Available Formats

PMRes 2023 4

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

PMRes 2023 4

Uploaded by

Copyright:

Available Formats

Institute

Prof. POUL ANDERSEN Prof. YORDAN GARBATOV Prof. JERZY MERKISZ

Prof. JIAHN-HORNG CHEN Prof. STANISLAW GUCMA Prof. VALERI NIEKRASOV

Computer-aided system for layout of fire hydrants

* Corresponding author: piotr.gomulka@pg.edu.pl (Piotr Gomułka)

Keywords: ship, fire hydrant, design, layout, particle swarm optimization

introduction FFS Resolution2 , classification society rules, and standards

4 POLISH MARITIME RESEARCH, No 4/2023

POLISH MARITIME RESEARCH, No 4/2023 5

6 POLISH MARITIME RESEARCH, No 4/2023

e 𝑖𝑖-th control point,

8 POLISH MARITIME RESEARCH, No 4/2023

POLISH MARITIME RESEARCH, No 4/2023 9

𝑓𝑓1 (𝑥𝑥) = ∑𝑛𝑛𝑎𝑎=1 𝐴𝐴𝑎𝑎 (𝑥𝑥) Fig. 6 Simulation

POLISH MARITIME RESEARCH, No 4/2023 11

12 POLISH MARITIME RESEARCH, No 4/2023

POLISH MARITIME RESEARCH, No 4/2023 13

14 POLISH MARITIME RESEARCH, No 4/2023

POLISH MARITIME RESEARCH, No 4/2023 15

13. D. Rutten, “Galapagos: On the logic and limitations of

14. C. Waibel, T. Wortmann, R. Evins, and J. Carmeliet,

15. J. M. Cichocka, A. Migalska, W. N. Browne, and E. Rodriguez,

16. T. Wortmann, “Opossum – Introducing and Evaluating

17. I. Anton and D. TĂnase, “Informed Geometries. Parametric

18. J. C. Lagarias, J. A. Reeds, M. H. Wright, and P. E. Wright,

19. J. Kennedy and R. Eberhart, “Particle Swarm Optimization,

20. L. M. Hiot et al., Handbook of Swarm Intelligence, vol. 8.

21. S. Bhattacharyya and P. Dutta, Handbook of research

22. G. O. Brown, “The history of the Darcy-Weisbach

23. K. Rudzki, P. Gomulka, and A. T. Hoang, “Optimization

16 POLISH MARITIME RESEARCH, No 4/2023

RETROFITTING THE BOW OF A GENERAL CARGO VESSEL

* Corresponding author: y.denev@tu-varna.bg (Yordan Denev)

introduction gas, modifying the bow design, or simply changing the

POLISH MARITIME RESEARCH, No 4/2023 17

where HFC is the𝑀𝑀𝑀𝑀𝐹𝐹𝐹𝐹𝐹𝐹𝐹𝐹

POLISH MARITIME RESEARCH, No 4/2023 19

Voyage parameters were selected in accordance with

Fig. 3. Hull form variant VAR2 with bulbous bow

Fig. 6. Voyage distance map

The distance of 3940 nm was travelled in 11 days, at an

Fig. 7. Voyage fuel cost and specific fuel consumption

Considering the main engine-specific consumption, ship

20 POLISH MARITIME RESEARCH, No 4/2023

Fig. 8. Energy Efficiency

POLISH MARITIME RESEARCH, No 4/2023 21

3. F. Technology, ‘Retrofitting a new bulbous bow’ [online,

DECLARATION OF COMPETING INTERESTS 10. International Maritime Organization, ‘Guidelines for

22 POLISH MARITIME RESEARCH, No 4/2023

14. A. Papanikolaou, ‘Ship Design- Methodology of

15. G. Butler, ‘Guide to Ship Repair Estimates’, Oxford, 2000,

16. ‘Made in China’ [online] available https://www.made-

17. ‘Steel Pipes and fittings, Shipbuilding steel price’

POLISH MARITIME RESEARCH, No 4/2023 23

USE OF THE AHP METHOD FOR PREFERENCE

* Corresponding author: jan.sierzputowski@pg.edu.pl (J. Sierzputowski)

INTRODUCTION safe at sea, economically efficient, and generally perform well

24 POLISH MARITIME RESEARCH, No 4/2023

POLISH MARITIME RESEARCH, No 4/2023 25

Step 1: Tabulate the output value Uʹ according to the input eʹand