Learning Aspects of Procedures for Ship Conceptual

Design Based on First Principles

Hernani L. Brinati, Oscar B. Augusto, Mardel B. de Conti
Department of Naval Architecture and Ocean Engineering – University of São Paulo
Av. Prof. Mello Moraes, 2231 – PNV/EPUSP –05508-900 – São Paulo (SP) – Brasil
[email protected], [email protected], [email protected]

Abstract – This paper presents the learning process, approaches, namely the point-based and set-based designs.
oriented to undergraduate students in Naval The following section - Models and Principles - presents
Architecture, of some procedures for the conceptual some models for the estimation of functional attributes of the
design of ships based on first principles. Firstly, solutions. Some of these models are mainly based on
considering the points of view of the different empirical methods, while others are more oriented to a first
stakeholders, the design requirements are identified and principles approach. Nevertheless, the whole frame of the
analyzed, taking into account the relevant physical, procedures is adequate to the first principles approach, and
economic, technological and social aspects. Based on this once the empirical models can be replaced and be analytical
information, merits and constraints are established in ones, the procedures can be adapted accordingly. The
order to follow a decision making process on the design following section describes a case study related to the
alternatives. An application example of this learning renewal of the fleet of an oil company. The last section
process in a ship design undergraduate course is presents conclusions about this learning process.
described, in which cost and risk are taken as design
merits. Certain relations among dimensions and inertias
are taken as descriptive parameters of a solution. The SHIP DESIGN PROCESS
bridge between merits and parameters is framed on the
following functionalities: cargo capacity, stability, In ship design there are many domain-specific models of the
resistance and propulsion, maneuverability, seakeeping, design process, but Evans’ design spiral is probably the most
strength. Genetic algorithms are used as a searching well known. This model emphasizes that many design issues
process. The scenario of the problem was chosen in interact and must be considered in sequence, in increased
accordance with the expectations of the students, being detail in each pass around the spiral, until a single design that
the renewal of the fleet of an oil transportation company. satisfies all constraints and balances all considerations is
As result, a form of product model is obtained, reached. This approach is essentially a point-based design,
specifically oriented to conceptual design, which since it leads to a single point in the space design. A
materializes the learning acquired by the students. disadvantage of this approach, as pointed out by Parsons [2],
is that it may not produce a global optimal solution.
Index Terms – Design based on first principles, parametric Nowadays, a different approach, taken from the automotive
design, ship design learning, ship design process. industry, is being used in the ship conceptual design. It is
called set-based design and, as a main feature, it defines
broad sets for the parameters’ design, in order to allow
INTRODUCTION concurrent design to begin, and keeps open these sets, so that
the design teams can see the difference in performance and
When the authors began planning the Ship Design course for cost among the different solutions.
the students of the 8th semester in Naval Architecture and The conceptual design process includes the following
Ocean Engineering, it was decided that a new learning phases: needs identification, requirements definition; design
process would be used. The course would provide the criteria selection and solutions framework development [3].
students the opportunity to develop a conceptual ship design Conceptual design influences the largest portion of the life-
in which they would sequentially identify the problem and cycle cost of the product, and thus the use of a set-based
analyze the relevant information, formulate relationships design approach is more appropriate to meet an optimal
between design parameters and functional attributes, and global solution.
consequently acquire a comprehensive discourse about the In searching for an optimal solution, it is not always
achieved solution, the principles that model the possible to make use of traditional prescriptive methods,
functionalities, and the searching process. Certainly it would which are often difficult to evolve to new designs; therefore
not be a simple task, since there was not enough time to alternative methods must be tried. In this regard, first
simulate the course development and also because all the principles methods are increasingly used [4], which means
past experience in this course had been with the use of that, as far as possible, analytical models are used to relate
Evans’ spiral design, apud Andrews [1]. functional attributes to design parameters. Based on these
The remainder of this paper is organized as follows. The attributes, merits are built in accordance to the design
next section - Ship Design Process - describes two different requirements in order to guide the optimization process.

Coimbra, Portugal September 3 – 7, 2007

International Conference on Engineering Education – ICEE 2007
 ∂A ∂V
A = (4 ⋅ L ) D = 1 ; V = L⋅B D =1
MODELS AND PRINCIPLES ∂D A 3 B +1 ∂D V
D D
4D
Weight Estimation and Balance
∂A
Again for B ≅ 1.8 , one has A ≅ 0,43 , showing that
Weight estimation can be done based on empirical D ∂D
regressions for structure, machinery, outfit and deadweight D
items other than cargo [5]. area change is about 43% of the volume change, given a
An additional approach can be used for estimating how specific change in height.
certain parameters such as length, breadth and depth have This type of analysis may give important directions on
influence on structural weight. The ship is considered as a design changes for adapting multiple compromises in the
longitudinal girder, which has the same length of the ship, early stages of the conceptual design. If one needs an
being its height is equal to the ship’s depth and its width increase of volume, for instance, probably the most effective
equal to the ship’s breadth. For a tanker with double bottom way to obtain it - constraining the analysis to changes in L, B
and double sides, one could consider for instance three and D – would be changing D, which would imply the least
flanges (for the double bottom and for the deck) and four impact on the steel quantity (proportional to area A), and so
webs (for both double sides). All the continuous longitudinal on.
material shall be considered as contributing to an equivalent Weight estimation can be done based on empirical
thickness for plating. regressions for structure, machinery, outfit and deadweight
The total area of the webs and flanges for the ship items other than cargo [5].
modeled as a girder would be:
Arrangement; Volume Estimation and Balance
A = 3⋅ L⋅ B + 4⋅ L ⋅ D ,
Arrangement in the conceptual design is here understood as
where L is the ship length, B the ship breadth, D the ship the definition of dimensions for peak tanks, cargo tanks,
height. This area would be roughly proportional to the cofferdams and machinery space, verifying whether these
structural weight of the ship. allocations are compatible with the ship’s main dimensions.
The total volume enclosed by the hull girder sides and If not, the main dimensions have to be modified so as to
bottom and deck would be: balance the needed and available volumes.

V = L⋅B⋅D , Static Stability

which would be roughly proportional to the cargo volume Ship’s transverse stability depends upon the metacentric
available. height, which is calculated based on the vertical position of
In the following, only relative increments or decrements the center of mass and of the center of buoyancy, as well as
of the structural weight and of the girder volume due to the metacentric radius.
relative changes in one of the main dimensions L, B or D An analytical model, based on a simplified geometric
will be considered. representation of the ship hull, possibly as a polyedric
Regarding ship length, one has: surface, could be built for the estimation of the position of
the buoyancy center and of the metacentric radius.
∂A ∂V Specifically, the metacentric height is calculated based on
A = (3 ⋅ B + 4 ⋅ D ) L = 1 ; V = B⋅D L =1 , the first moment of area of the ship design waterplane. As for
∂L A ∂L V
L L the center of mass calculation, the vertical position of the
main weight items must be estimated beforehand.
showing that volume and area changes are equal to a specific
change in length.
Primary Structure
Regarding ship breadth, one has:
The main role of any ship structure is to support loads – its
∂A ∂V own weight, cargoes and sea environment – and deflect as a
A = (3 ⋅ L ) B = 1 ; V = L⋅D B =1
∂B long flotation beam, generally referred as a hull girder
∂B A 4 B
V
B 1+ 3 primary structure. Such a hull girder is composed of by
B several panels, mainly portions of shell plating reinforced by
D
longitudinal stringers and transverse frames, being the last
For B ≅ 1.8 (as for a tanker, for instance), one has ones responsible to keep the hull girder cross section in its
D original shape. The hull girder has to support vertical and
∂A horizontal bending moments and shear forces. Access to
A ≅ 0,57 , showing that area change is about 57% of the
∂B details of the main hull cross section is only available after
B
several design cycles, when the hull forms are detailed.
volume change, for a specific change in breadth.
Nevertheless just the ship’s lightweight is used in the early
Regarding ship depth, one has:

Coimbra, Portugal September 3 – 7, 2007

International Conference on Engineering Education – ICEE 2007
stages of ship design. To obtain details of the main hull cross Layout diagrams provided by engine builders are used to
section, it is necessary to have all loads and a first estimate select feasible engines, with the requirement that the point
for the necessary longitudinal material distribution for the representing the propeller demand on speed and power lies in
main hull. At the beginning of the design process, one may the layout diagram. In order to do so a power margin is
use classification societies’ rules to obtain the loads, the defined to evaluate the installed power, this corresponding to
minimum hull girder bending moment and shear forces and the specified maximum continuous power.
primary minimum safety factor. Then one may use the
classical beam theory to deal with a hull girder cross section
Maneuverability
composed of by plates with equivalent thickness necessary to
take into account for the longitudinal reinforcements. Using Some maneuverability models are based on fundamental
such methodology, a double hull tanker may be composed of principles, but they have some parts or parameters not
one to six equivalent thicknesses, depending on the detail completely developed or identified, estimated in a way that
level adopted at the early design stages. These thicknesses fit some set of empirical and/or experimental data according
refer to the double bottom, double shell, main deck and to some approximation criteria. These models are usually
longitudinal bulkhead, if present. specific for certain types of ships and maneuvers. [7] and [8]
are examples of this approach. Other models rely more
extensively on fundamental principles, though with some
Resistance and Propulsion
limitations such as the representation, for instance, linearity.
The mathematical model provides not only an estimation of [9] and [10] are examples of this approach.
the hull resistance and engine power demand but also leads All these models are quite convenient for application in
to the selection of the propeller engine set for a low-speed the early stages of ship design, since the involved parameters
Diesel power plant, which is the most common type for are generally some overall dimensions, as well as some
tankers. shape and functional coefficients of the hull and appendices,
Although an analytical model could be used to evaluate which could easily be taken as conceptual design parameters
one component of the hull resistance, namely the wave in an optimization search process.
resistance, Holtrop’s formulation [6] is adopted to evaluate Attributes like those that are considered as interim
the total resistance, since for tankers the frictional resistance maneuver criteria by the International Maritime Organization
is, at least, equal to fifty percent of the total resistance. [11] can be evaluated through these methods, for instance:
Holtrop’s formulation is based on a statistical analysis of • Turning ability
resistance data. A resistance service margin is included to • Initial turning ability
provide the added power required in order to overcome in • Yaw checking and course keeping ability
service the added resistance from hull fouling, waves and • Stopping ability
wind effects. Additionally to the ship maneuvering in deep
Holtrop’s models are also used to predict values for unrestricted waters, models for maneuvering in restricted
wake coefficient and resistance increase (thrust deduction) waters and for organizational and human factors, influencing
factor, as well as to provide the relative rotative efficiency. ship navigation and steering, should be considered since the
In order to select propeller alternatives, number of early stages of ship design, due to their potential impacts on
blades, disk area coefficient (expanded blade area ratio) area ship safety. In fact, as pointed out in [12], “including
ratio and pitch-diameter ratio are used as design parameters. restricted waterway maneuverability as an important spoke
The maximum propeller diameter, which usually provides in the ship design spiral would seem a necessary step to
best efficiency, is assumed, but other values can be used as enabling proper tradeoffs in vessel design”. An approach
long as engine speed constraints make it necessary. which considers indirectly maneuverability qualities in
Polynomial representations of dimensionless propeller torque restricted waters is addressed in the section “risk evaluation”.
and thrust coefficients for Wageningen B Screw Series
propellers [6] are used in order to perform the hull-propeller-
engine interaction. Keller’s cavitation criterion [6] is used to Seakeeping
impose an external cavitation constraint since charts for open Seakeeping is here restricted to a one degree-of-freedom
water tests do not take it into account. model of the ship rolling motion. The problem is subdivided
Essentially, the determination of the best propeller- in three cases: 1) hull free to roll in previous undisturbed
engine set consists in an optimization problem in which the waters; 2) roll motion in regular waves; 3) roll motion in
selected propeller has to produce a thrust equal to the irregular waves. In the first case, there are three terms,
“augmented” ship resistance and a Diesel engine that meets namely, the virtual inertia (mass inertia plus added inertia),
the demand of propeller speed and power. The propeller the damping and the hydrostatic restoration. This problem is
speed is obtained from the propeller advance coefficient, typically nonlinear in the damping, which usually is
given by the force equation. The propeller torque coefficient represented by an odd polynomial [13]. As for a linear plus
is also obtained from the value of the advance coefficient, parabolic representation for the damping, the empirical
combined with the estimate of the relative rotative and formulation compiled in [14] may be used, with the approach
transmission efficiencies, to evaluate the required engine of an equivalent linear representation, in the sense that both
power. representations lead to the same energy dissipation per cycle.
As for the restoration term, it can be represented based on the

Coimbra, Portugal September 3 – 7, 2007

International Conference on Engineering Education – ICEE 2007
metacentric height. The added mass depends on the the design process is a difficult task. The best hull form will
geometry of the cross sections and has to be estimated conflict with the requirements for cargo spaces, for example.
accordingly. This step enables to estimate the roll natural Usually, the designer has conflicting objectives to satisfy.
period and decay in previously still waters. In the second Before the advent of multi-objective optimization
case (forced motions under regular waves), the exciting techniques, multi-objective problems were addressed by
moment is estimated quasi-hydrostatically considering a collapsing all the objectives into a single objective. A classic
mean wave slope [15]. This step leads to the response (i.e., single-objective) optimization algorithm was then used
amplitude operator of the ship roll motion. The third case to minimize (or maximize) this collapsed objective.
(ship roll in irregular waves) is based on the superposition However, one needed to decide a priori as how to prioritize
assumption of the real sea excitation according to [16], one objective in detriment of others before knowing the
leading to the roll significant height. resulting alternatives. The key advantage of multi-objective
optimization is that it does not require the user to make
premature decisions about the ideal trade-off. One approach
Cost Evaluation
to handle multi-objective design problems is to employ the
Cost evaluation may be done based on weight items, such as concept of Pareto optimality. Pareto optimality was
steel, outfit, deadweight item other than cargo, etc., as well introduced in the late eighteen hundreds by the economist
as on power demands. The weight items and power are used Vilfredo Pareto. A solution is said to be Pareto optimal if
in regressions enabling the estimation of acquisition and there exists no other solution that is better in all attributes.
operational costs. This implies that, in order to achieve a better value in one
objective, at least one of the other objectives is going to
deteriorate if the solution is Pareto optimal. Thus, the
Risk Evaluation outcome of a Pareto optimization is not one optimal point,
Risk is here restricted to environmental impacts due to oil but a set of Pareto optimal solutions that visualize the trade-
spills caused by hull perforation in groundings and off between the objectives. One of the most used types of
collisions. algorithms in multi-objective problems is referred as Genetic
Ref. [17] considers the powered and drift groundings of Algorithms (GAs). GAs have appeared in recent years, and
ships. The “technique for human error rate prediction” are so called because they try to simulate, and replicate, the
(THERP) is used to predict human error rates based on mechanisms of natural selection and genetics. They search
“possible human task activities and the corresponding error for the absolute maximum of a function being optimized, (or
probabilities”. It is used in connection with empirical data the group of maxima in the case of multi-objective
concerning “human error probabilities” (HEPs) as modified optimization) by ‘mating’ designs already assessed, on the
by “performance shaping factors” (PSFs). Fault trees and basis of probabilities geared to their objective values. GAs,
event trees lead to the identification of system failures and may be specifically tailored to the problem at hand, are
sequences, whose associated probabilities can be assigned simple enough to be easily understood and accepted among
using the THERP and statistical data. The system can be the students, and are robust and reliable enough for most
further modified to meet the safety requirements concerning purposes. There are several alternative methods for the
grounding probability level. selection of the designs which will ‘mate’. GAs also provide
Collision can be treated in a similar way to powered a further mechanism, whose aim is to keep the possibility of
grounding. diversity in the population of designs in which is modeled:
Mean oil outflow can be evaluated, given that collision the ‘mutation’. It is now recognized that algorithms which
or grounding occurred, using historical data information belong to the GA’s class are among the best ones which lead
about position and penetration of damages. A procedure for to effective multi-objective optimization. It is also known
oil spill from bunker tanks was proposed to IMO [18]; a that, from an engineering perspective, they are very robust,
similar procedure was used in [19]. since they provide the possibility to find a better solution,
In the early stages of design, the definition of double and hopefully also the maximum extreme, of the function to
side width and double bottom height is done simultaneously be optimized. Conversely, for ‘classical’ methods, their
with the definition of the main dimensions of the ship, the success relies heavily on the proper, and usually far from
lengths of cargo and peak tanks, the length of machinery easy, choice of the initial condition. In other words, genetic
spaces and cofferdams, etc. Altogether, the ship volumes algorithms always provide, if not the ‘best’ solution, a ‘good’
must be compatible with the allocation of cargo and ship solution. The claimed drawback of genetic algorithms is their
systems; weights and displacement must also be balanced. In low convergence rate and high computational costs, as a
this process of volumes and weights balancing, it becomes consequence of the excessive number of evaluations needed
clear whether or not there is a margin for increasing the for the objective-functions and constraints. However, this
double side width and double bottom height, as well as their becomes negligible if the entire design process is viewed,
impact on the mean oil spill. and managed, from an engineering perspective.

Optimization process
CASE STUDY: AN ILLUSTRATIVE EXAMPLE
Considering that a ship a complex product involving many
engineering branches, to set up a simple objective to guide
Context and Motivation

Coimbra, Portugal September 3 – 7, 2007

International Conference on Engineering Education – ICEE 2007
A Brazilian oil company is renewing its ship fleet demanding being the discussion of the process itself, more than the
five different types of ships: gas carriers, product carriers, results, an important part of the learning process.
Panamax, Aframax and Suezmax. The basic ships
specification data were available from the company’s site TABLE I
RESULTS FOR THE HANDYMAX LIGHT PRODUCTS CARRIERS
and were used as a starting point for the students’ design.
Group / Parameter Group A Group B
The students were enrolled in a one semester course in
Length (m) 183.4 186.1
ship design. The class summed up of about 60 students; 15 Breadth (m) 32.2 32.1
groups were formed, each three groups working with the Design draught (m) 12.1 12.4
same ship type. The authors were lecturers in this course and Depth (m) 16.8 17.2
Displacement (ton) 60,232 57,280
each one advised five groups.
Double bottom
In the beginning of the term the students already knew height (m) 2.0 2.2
the traditional ship design process based on the spiral model. Double side width
When the parametric design paradigm was introduced, the (m) 2.0 2.0
Number of tanks 12 12
students showed some resistance to follow this procedure.
This was considered understandable, since in the spiral
process, an effective solution is created, meaning that a TABLE II
specific ship is developed, while in the parametric process RESULTS FOR THE SUEZMAX CRUDE OIL CARRIERS
what is built is a set of attribute estimation models, not Group / Parameter Group C Group D
Length (m) 264.9 265.9
enabling the students to visualize specific solutions. Because, Breadth (m) 49.6 46.8
in the parametric process, it is possible to consider various Design draught (m) 17.0 17.0
set points for design, the specification of each solution Depth (m) 21.4 26.8
remains open until the convergence of the selection Displacement (ton) 194,600 188,170
Double bottom
procedure. height (m) 2.3 3.2
However, as the students continued working on their Double side width
projects, they became aware of the strength of the parametric (m) 3.5 4.0
design procedure and finally engaged themselves in it. Number of tanks 12 12

Results
Learning aspects
Tables I and II present the main dimensions of a couple of
final solutions for two types of ships: Handymax Light The students were free to choose any computational
Products Carrier and Suezmax Crude Oil Carrier. language and platform to develop their computational design
The requirement for cargo volume for the Handymax systems. All of them have reached the main purpose and
was 54000 cubic meters with a mean density of 0.85 tons per some have mastered their abilities in systems development,
cubic meter. For the Suezmax, the requirement was 1050 arriving at interactive design systems. Others, based on the
thousands of barrels with a mean density of 0.94 tons per genetic algorithm basic principles, have got a deep
cubic meter. The minimum service speed was required to be understanding of the metaheuristic, so that they could
15 knots. develop new approaches.
Group B ship has a smaller block coefficient than Group On one hand, with the classical spiral based design,
A ship. Despite the fact that Group B ship has higher values there is more opportunity for the student to go deeper in
of double bottom height, depth and length, its light weight some ship’s systems details. On the other hand, parametric
resulted smaller than that of Group A ship. design allows the consideration of a large number of
Group D ship has also a lighter weight than Group C alternatives for design solutions, giving the students a good
ship, and this is understandable due to its smaller breadth, perception of the influence of each design parameter on the
despite its larger depth, double bottom height and double solution attributes.
side width.
Ships of Groups B and D seem to have a better design
than the ships of Groups A and C, respectively, due to the CONCLUSIONS
smaller displacement and larger double bottom height and /
or double side width. This is though questionable, since For each class of solutions characterized by its typical
different models have been used for the ship attributes geometries and technologies, a bridge was built, filling the
estimation, as well as different optimization objective gap between the merits / constraints and the parameters
functions and constraints have been assumed. which describe the solutions of that class. This bridge is in
Nevertheless, all of these models and assumptions about fact an algorithm that, given the descriptive parameters of
objectives and constraints, as well as the connection of the each solution alternative, enables the calculation of the
models and adopted assumptions to reality and the merits and the enforcement of the constraints. The algorithm
effectiveness of the search process, are explicit and prone to is structured on the basis of first principles supplemented by
critical assessment by the various groups and advisors. This empirical observations, and is applied in a search for the best
is a very significant advantage of a more rational oriented solution that verifies the applicable constraints.
design process in respect to a more empirical procedure,

Coimbra, Portugal September 3 – 7, 2007

International Conference on Engineering Education – ICEE 2007
 This learning process enabled the students to: a) become [6] Lewis, E. V. (editor), “Principles of Naval Architecture”, Second
Revision, The Society of Naval Architecture and Marine Engineers,
aware of the problem that was being considered, identifying
New Jersey, vol. II (Resistance, Propulsion and Vibration), 1988.
and analyzing the relevant information; b) make use of first
principles, as well as empirical data, to conceive classes of [7] Inoue, S.; Hirano, M.; Kijima, K., “Hydrodynamic derivatives on ship
maneuvering”, International Shipbuilding Progress, Vol. 28, No #
solutions to the problem, developing or reinforcing their 321, May/1981, pp. 112-125.
conceptual understanding regarding the relations between
geometric parameters and functional attributes, which lead to [8] Takashima, J., “Ship manoeuvering motion due to tugboats and its
mathematical model”, Journal of the Society of Naval Architects of
the merits; c) acquire an explicative discourse about the Japan, No # 160, 1986, pp. 93-104.
achieved solution and how it works, the algorithm and the
[9] Clarke, D.; Gedling, P.; Hine, G., “The application of maneuvering
principles that model functionalities, the process for criteria in hull design using linear theory”. Transactions of RINA, Vol.
searching for the best solution, and about the design process 125, 1983, pp. 45-68.
itself.
[10] Clarke, D.; Yap, M. L. R., “A multi-criteria view of the IMO
manoeuvring requirements”, Transactions of RINA, Vol. 143, 2001,
pp. 99-111.
ACKNOWLEDGMENT [11] 23rd International Towing Tank Conference, “Group Discussion B.2:
IMO Standards and ITTC (Particularly related to Manoeuvring and
The authors are indebted to the students who conducted the Capsizing)”, Proceedings of the 23rd ITTC, Vol. III, 2002.
projects of ship design for their engagement and contribution [12] Gray, W. O.; Waters, J.; Blume, A.; Landsburg, A. C., “Channel
to certain aspects of modeling and estimation processes. Design and Vessel Maneuverability - Next Steps”, Society of Naval
Architects and Marine Engineers, 2002.
[13] Taylan, M., “The Effect of Nonlinear Damping and Restoring in Ship
REFERENCES Rolling”, Ocean Engineering, Vol. 27, pp. 921–932, 2000.
[14] Chakrabarti, S., “Empirical Calculation of Roll Damping for Ships and
[1] Andrews, D. J., “A Comprehensive Methodology for the Design of Barges”, Ocean Engineering, V. 28, 2001, p. 915–932.
Ships (And Other Complex Systems)”, Proceedings of the Royal
Society, London, Vol. 454, pp. 187-211, 1998. [15] Zaojian, Z., “Ship Manoeuvring and Seakeeping”, Shanghai Jiao Tong
University, Lecture Notes, Available in [http://naoce.sjtu.edu.cn],
[2] Parsons, M., G., “Parametric Design, in Ship Design and 2006.
Construction”, Edited by Lamb, T., The Society of Naval Architecture
and Marine Engineers, New Jersey, pp 11-1 to 11-48, 2003. [16] St Denis, M.; Pierson, W. J., “On the Motion of Ships in Confused
Seas”, Transactions of the SNAME, 1953.
[3] Chou, Y.-C.; Jen, F., “Applying Neural Networks In Quality Function
Deployment Process For Conceptual Design”, Journal of the Chinese [17] Amrozowicz, M.; Brown, A.J.; Golay, M., "A Probabilistic Analysis
Institute of Industrial Engineers, Vol. 21, No. 6, 2004, p. 587-596. of Tanker Groundings", 7th International Offshore and Polar
Engineering Conference, Honolulu, Hawaii, May, 1997.
[4] Committee IV.1 of the 16th International Ship and Offshore Structures
Congress, “Design Principles And Criteria”, Proceedings of the 16th [18] INTERTANKO, Evaluation of Accidental Oil Spills from Bunker
International Ship and Offshore Structures Congress, V. 1, 20-25 Tanks”, Available from <www.sname.org/committees/tech_ops/
August 2006. bunkertank/paper-IMO-DE.pdf>.
[5] Watson, D. G. M., and Gilfillan, A. W., “Some Ship Design Methods”, [19] Lützen, M.; Simonsen, B. C., “Grounding Damage to Conventional
Transactions RINA, Vol. 119, 1977. Vessels”, Available from < http://www.skibstekniskselskab.dk/
download/WMTC/F17(R97).pdf>.

Coimbra, Portugal September 3 – 7, 2007

International Conference on Engineering Education – ICEE 2007