A consortium of maritime classification
societies and university research groups has developed an improved
set of reliability-based rules for the structural design of ships.
Unlike traditional rules based on empirical safety factors, the new
rules are derived from detailed analysis of the uncertainties in the
computer models used to design the hulls of ships, notably models
for loading effects and structural strength. Already, a number of
societies are considering the SHIPREL rules.
When we board a ship, we would like to
think that the vessel is strong enough to withstand the sea conditions
that may be encountered on the voyage. And so it is - designers
always allow generous safety factors when calculating the structural
strength of ships.
However, there is a modern approach to quantify structural safety.
In principle, one could calculate the probability of a hull failing
under the expected range of conditions at sea, and then design the
ship so that the chances of failure were smaller than some minimum,
acceptable value. In civil engineering, for example, designers have
been using such reliability based methods since the late 1970s to
replace traditional safety factors in the design codes. For each
structure, such as a bridge, the probability of failure can be calculated
during the planning stage and the design modified accordingly. This
ensures that the bridge is safe without incurring unnecessary expense.
Unfortunately, the models currently used to calculate the strengths
of ships' hulls, and the stresses encountered in operation, have
not been sophisticated enough to allow reliability-based methods
to be adopted by the world's classification societies. Established
by the insurance industry to improve the safety of shipping, the
societies lay down rules for the design of different kinds of vessels.
In SHIPREL, ship classification societies in France, Germany and
Italy, together with university research groups in Portugal and
Denmark, joined forces to develop new reliability-based methods
for the structural design of ships.
Modern ships come in a vast range of shapes and sizes, so the partners
decided to concentrate on just two types: oil tankers and container
vessels. These represent opposite ends of a spectrum of characteristics,
broad and slow on one end, and narrow and fast on the other.
Reliability analysis requires answers to many difficult questions
about the nature of hull failure, the range of conditions expected
at sea, the ultimate strength of the hull, and so on. The project
concentrated on the primary strength of ships; that is, the strength
of the hull under longitudinal bending and, in particular, the design
of the midship section where stresses and strains are greatest.
Stresses and strains
The first part of the project was to develop new computer models
to represent the stresses in the structures of a ship's hull. These
stresses have several sources. First, stresses arise from the loading
of the ship - the way cargo is distributed within it. These still
water effects will change during a voyage, as cargoes are loaded
and unloaded; there are large differences, for example, between
full and empty oil tankers. As well as searching existing databases
for loading information on tankers, SHIPREL has collected new data
from voyages of 40 container ships to quantify the effects of different
A second source of stress, also tackled by SHIPREL, is a variety
of effects caused by the motion of the sea waves. These range from
gentle rolling to violent slamming, and vibrations course through
the ship due to the impact of the waves. The partners developed
an improved method for predicting the effects of large waves.
To test the models developed by SHIPREL, the partners fitted a container
ship with instruments to record stresses, strains and temperatures.
The ship was monitored for two years as it sailed the Atlantic between
France and North America.
Modelling of structural strength
The models used at present for calculating the structural strength
of ships suffer from many uncertainties. Ships are complex structures
and the models have to be simplified to enable computers to handle
In finite-element analysis, a well-established but complex method,
the actual modelling process is time-consuming, requiring skilled
and expensive personnel, so detailed modelling is often restricted
to the midship section rather than the whole hull. A cheaper and
simpler method is to represent the entire hull as a single beam
subject to bending and twisting forces, but it is uncertain how
closely this model represents a real ship. Also, while the distribution
of wave-induced forces at each point on the hull should be taken
into account, it is usually simplified for ease of computation.
Whichever method is used, designers must take care with interpreting
the results - the calculated strengths may be very different to
those of a real ship.
To get some idea of the practical uncertainties, and thus estimate
the safety factors, the participants made several calculations using
different models of the same ship. They also looked at the effects
of fatigue on the strength of welded joints.
One notable success is a simplified procedure for calculating the
collapse strength of a girder consisting of stiffened panels. The
new method takes only a few seconds of computer time compared with
several hours for non-linear finite element methods.
New rules for ship design
At the end of a long and complex series of computations, the SHIPREL
team produced simplified procedures and formulae for use in ship
design. By combining the uncertainties in the loading models and
the strength models they have produced a set of safety factors based
upon sound engineering principles. They propose that ships be designed
so that there is one chance in 10,000 of a structural failure occurring
in any one year.
The new rules will also be helpful in assessing the condition of
existing ships and estimating how many years they can safely continue
SHIPREL is arousing great interest in the world of shipping. Many
research publications have stemmed from the work, and a special
issue of the Marine Structures journal has been devoted to it. In
autumn 1996 a presentation of the project was made to the Society
of Naval Architects and Marine Engineers in New York.
Some classification societies are in the process of modifying their
design rules in the light of the SHIPREL findings. But while the
demonstration of reliability methods is of great interest to all
those concerned with the safety of ships, the various software modules
devised by SHIPREL are not yet at the stage where they can be used
for routine engineering. A single, integrated software package for
designers is the next step towards more reliable structures in the