THE SECOND GENERATION OF INTACT STABILITY CRITERIA:

AN OVERVIEW OF DEVELOPMENT
W. Peters (M), V. Belenky (M), C. Bassler (M), K. Spyrou (M), N. Umeda (M), G. Bulian (V), B. Altmayer (V)
This paper describes the current state of development of the second generation of International Maritime
Organization (IMO) intact stability criteria. The objective of the paper is three-fold. First, it is a brief review of the
recent history of related IMO development with emphasis on the question: why is the second generation of stability
criteria needed? Second, the paper describes the framework of the new criteria in which they are complementary to
the existing IMO criteria. The new criteria are applied only to those designs that may be susceptible to one of the
modes of stability failure considered (parametric roll, pure loss of stability, surf-riding/ broaching). This is achieved
through a set of formalized procedures of vulnerability checks. The ideas, proposals, and justifications of these
procedures represent the novel contents of the paper. Last, the paper reviews available technologies for development
of direct stability assessment methods or performance-based criteria that should be available for those rare cases
when susceptibility to these modes of failures is too high.
INTRODUCTION: REVIEW OF INTACT STABILITY DEVELOPMENTS AT IMO
The origin of the first-generation intact stability criteria that are included in the foundation of the International Code on Intact Stability,
the 2008 IS Code (IMO, 2009), can be traced to the pioneering works of Rahola (1939), as well as early versions of the weather
criterion developed in the 1950s. The history of development and the background of these criteria are described by Kobylinski &
Kastner (2003); a summary of the origin of these criteria is also available in Chapter 3 of the Explanatory Notes to the International
Code on Intact Stability (MSC.1/Circ.12811).
The first generation intact stability criteria were originally codified at IMO in 1993 as a set of recommendations in Res. A.749(18) by
taking into account, among other things, former Res. A.167(ES.IV) (“Recommendation on intact stability of passenger and cargo ships
under 100 metres in length” which contained statistical criteria, heel due to passenger crowding, and heel due to turn, 1968) and Res.
A.562(14) (“Recommendation on a severe wind and rolling criterion (Weather Criterion) for the intact stability of passenger and cargo
ships of 24 metres in length and over,” 1985). These criteria were codified in the 2008 IS Code and became effective as a part of both
the SOLAS and International Load Line Conventions in 2010 (IMO Res. MSC.269(85) and MSC.270(85)).
The criteria in Part A of the 2008 IS Code are based on a traditional empirical/statistical approach, with the exception of the Weather
Criterion. The criteria for passenger vessels associated with heeling due to turn and the crowding of passengers to one side are
formulated using a physics-based mathematical model of ship heeling.
The Weather Criterion is based upon a mathematical model of a ship heeling under the action of a sudden wind gust after being excited
by regular waves and a steady wind. The parameters of the Weather Criterion were “tuned” using a sample population of ships, which
limits the applicability of the Weather Criterion, in addition to the assumptions of the mathematical model used. In response to this, an
alternative experimental approach for the Weather Criterion was also adopted by IMO (cf. MSC.1/Circ.1200 and MSC.1/Circ.1227).
The introduction of ships with characteristics and/or modes of operation which are significantly different from the reference population
of ships on which the first generation criteria were based challenges the assumption that adequate intact stability is provided using the
current criteria. A series of stability-related accidents in the last 15 years, involving ships such as the APL China, M/V Aratere, and
Chicago Express, clearly demonstrates that intact stability criteria must be revisited.
The development of the second generation intact stability criteria started in 2002 with the re-establishment of the intact-stability
working group by IMO’s Subcommittee on Stability and Load Lines and on Fishing Vessels Safety (SLF) (cf. Francescutto, 2004,
2007). However, due mainly to the priority of revising the IS Code for approval, the actual work on the second generation intact
stability criteria did not commence in earnest until the 48th session of the SLF in September 2005. The Working Group decided that
the second generation intact stability criteria should be performance-based and address three modes of stability failure (SLF 48/21,
paragraph 4.18):
 Restoring arm variation problems, such as parametric excitation and pure loss of stability;
 Stability under dead ship condition, as defined by SOLAS regulation II-1/3-8; and
 Maneuvering related problems in waves, such as broaching-to.
1
References to IMO documents such as “MSC.1/Circ.1281” appear in the list of references with an “IMO” prefix, i.e. as: IMO
MSC.1/Circ.1281. As there is no ambiguity in the names of the IMO citations, the year will be omitted from the citations.

The Second Generation of Intact Stability Criteria: An Overview of Development

A similar formulation was included in the preamble of the 2008 IS Code, indicating the direction of long-term development. However,
the restoring arm variation problem was considered as two problems: modes of parametric roll and pure loss of stability; hence, four
stability failure modes were considered.

During this initial development, there was general agreement that the second generation criteria should be based on the physics of the
specific phenomena leading to stability failure. The design and modes of operation of new ships take on characteristics that cannot,
with confidence, rely solely on the statistics of failures and regression-based methods. Also, there was general agreement on the
desirability of relating the new criteria to probability, or some other measures of the likelihood of stability failure, as methods of risk
analysis have gained greater acceptance and become standard tools in other industries (e.g. SLF 48/4/12).

These considerations led to the formulation of the framework for the second generation intact stability criteria, described in SLF 50/4/4
and discussed at the 50th session of SLF (May 2007). The key elements of this framework were the distinction between performance-
based and parametric criteria, and between probabilistic and deterministic criteria. Special attention was paid to probabilistic criteria;
the existence of the problem of rarity was recognized for the first time and a definition was offered. Also, due to the rarity of stability
failures, the evaluation of the probability of failure with numerical tools was recognized as a significant challenge.

By that time (2007), there was already some experience in the maritime industry on how to handle some issues related to dynamic
stability. Following a parametric roll accident on the APL China (France et al., 2003), the American Bureau of Shipping (ABS) issued
a guide for the assessment of parametric roll for containerships (ABS, 2004). The guide was based on a multi-tiered assessment
procedure. The first level—the susceptibility criteria— was built upon evaluation of changing GM in regular waves and the Mathieu
equation. If the ship was found to be susceptible to parametric roll, then a more complex criterion was applied. This “severity”
criterion involved the calculation of the full GZ curve in waves and numerical integration of the roll equation. If the roll response was
“severe enough,” then advanced numerical simulations were applied and ship-specific operational guidance was developed using a pro-
gram such as the Large Amplitude Motion Program (LAMP) (Lin & Yue, 1990). While conservative, the susceptibility and severity
criteria were still capable of identifying ships for which parametric roll was not possible.

Also at that time, the work of Germanischer Lloyd was focused on numerical assessment procedures using the advanced numerical
code GL Simbel (Brunswig & Pereira, 2006, Shigunov & Pereira, 2009). Furthermore, the Germanischer Lloyd development was
focused on the preparation of ship-specific operational guidance for the avoidance of parametric roll (Shigunov, 2009).

Besides the efforts by classification societies, significant progress was achieved in developing training programs in order to increase
crew awareness of parametric roll. An instructional video, produced by Herbert Engineering Corporation, is one successful example of
this activity1.

Analysis of these experiences led to an understanding that a multi-tiered approach should be applied for the development of the second
generation intact stability criteria as a way to avoid unnecessary work. In view of this, the idea of vulnerability criteria was first
formulated in the paper by Belenky et al. (2008). The paper also gave a broad review of the physical background of the modes of
dynamic stability failure considered. By virtue of its greater detail, this paper provided “explanatory notes” to SLF 50/4/4 and was
submitted to the 51st session of SLF (SLF 51/INF.4) as additional information.

The framework of the second generation intact stability criteria took shape based on the work of the intersessional correspondence
group (SLF 51/4/1 Annex 2). This document formalized the concepts contained in SLF 50/4/4; in particular, a clear distinction was
made between a criterion and a standard, the former being “an instrument of judging,” while the latter is a boundary between
acceptable and unacceptable.

In 2005, the Japan Society of Naval Architects and Ocean Engineers (JASNAOE) established a Strategic Research Committee on
Estimation Methods for Capsizing Risk for the IMO New Generation Stability Criteria (SCAPE Committee). The outcome of this
program was reported in five sessions of JASNAOE; some other results were reported in English at the Osaka Colloquium (Ikeda, et
al., 2008). An overview of this work is available from SLF 51/INF.6. In the meantime, certain developments in the field were affected
by the increasing consideration and practical formulation of the so-called “critical wave groups” approach. This was used for
probabilistic intact stability assessment during the European SAFEDOR project (e.g. Themelis & Spyrou, 2007), which allowed for a
practical interface between the deterministic and probabilistic viewpoints. SNAME established a Dynamic Stability Task Group whose
purpose is to provide a detailed review of developments in the field of dynamic stability (SLF 53/3/3).

Vulnerability criteria were the focus of the 1st and 2nd International Workshops on Dynamic Stability Consideration in Ship Design
(DSCSD) (Kobylinski, 2009).

1
Trailer available from www.herbert.com/videos/ParametricRoll/

The Second Generation of Intact Stability Criteria: An Overview of Development

The development of the second generation intact stability criteria was intensively discussed during the 10th International Conference on
Stability of Ship and Ocean Vehicles and in the following 11th and 12th International Ship Stability Workshops (Degtyarev, 2009; van
Walree, 2010; Belenky, 2011). In particular, a review was presented that examined the suitability of methods for vulnerability criteria
(Bassler, et al., 2009).

The consideration of excessive accelerations was also recently added to the list of stability failure modes (SLF 53/19, paragraph 3.28)
following the partial stability failure of Chicago Express, which resulted in crew injuries and loss of life (BSU, 2009). While
technically this issue is well-known, it has not yet been included in the regulatory framework. In a recent study, Shigunov et al. (2011)
considered a vulnerability check for excessive accelerations based on initial GM and roll damping.

These discussions and developments were formed into proposals presented and discussed at the 52nd and 53rd sessions of SLF (cf. SLF
52/3/1, SLF 52/INF.2, SLF 53/3/1, SLF 53/3/7, SLF 53/3/8, SLF53/3/9, SLF 53/INF.8, SLF 53/INF.10).

THE CONCEPT OF THE SECOND GENERATION OF INTACT STABILITY CRITERIA

The assessment of dynamic stability in realistic wave conditions is a formidable task. The 2008 IS Code recognized this:

“In particular, the safety of a ship in a seaway involves complex hydrodynamic phenomena which up to now have not
been fully investigated and understood. Motion of ships in a seaway should be treated as a dynamical system and
relationships between ship and environmental conditions such as wave and wind excitations are recognized as
extremely important elements. Based on hydrodynamic aspects and stability analysis of a ship in a seaway, stability
criteria development poses complex problems that require further research.”

A realistic seaway is also random, i.e. described as a stochastic process; therefore, the problem must also be considered in a
probabilistic context. Further, the occurrence of stability failure (a random event) is rare—such that the natural period of roll can be
considered as infinitely small in comparison with the expected time before such an event.

The existence of the problem of rarity makes direct, brute-force numerical simulation impractical for the evaluation of dynamic
stability failure. There are special procedures that allow the problem of rarity to be addressed, which are discussed later in this paper.
However, the solution may include numerical simulations using hydrodynamic codes (such as LAMP 1, FREDYN 2, GL-SIMBEL 3,
TEMPEST 4, and others), hybrid codes (such as CAPSIM 5, LAIDYN 6, SHIXDOF 7, SIMCAP 8, and others) or simpler ordinary
differential equation (ODE) tools (such as OU BROACH 9, ROLLS 10, and others). Preparing input data for these tools may include
model testing. While these tools represent the current state of the art (cf. Beck & Reed, 2001), their application is expensive and
requires proper justification of the necessity of their application, because not all ships are vulnerable to these stability failures.

This justification can be completed in the form of a multi-tiered approach (see Figure 1), whereby a ship would be checked for
vulnerability in the first tiers and, if found vulnerable, then the ship would be evaluated using state-of-the art direct stability assessment
methods. Taking into account the intended regulatory application, two tiers of vulnerability criteria would be applied. The first level is
meant to be very simple and conservative. Its main purpose is to distinguish ships (and the loading conditions) that are clearly not
vulnerable to a given stability failure mode from those that, in principle, may be. Because further analysis of the vessels that are not
vulnerable would be redundant, the cost of performing such further analysis should be avoided.

As an example, very large crude carriers are wall-sided for most of the length of the hull and, therefore, cannot experience significant
stability changes in waves. Therefore, this type of ship is not expected to be vulnerable to the righting-lever variation problems of
either parametric roll or pure loss of stability. By basing the first-level vulnerability criteria on the geometry characteristics of the hull
and speed, rather than the ship type per se, an easy first assessment of vulnerability may be made. Therefore, the criteria will remain
valid for any novel ship design.

Because the level-one criteria are to be simple and conservative, some occasional “false positives” may be expected. Again, to reduce
the time and cost of stability assessment, a second level of vulnerability criteria is introduced. The second level is meant to be less
conservative than the first, based on simplified physics and involving calculations with reduced computational efforts and
straightforward applications following suitable guidelines.

There has been very active development of the vulnerability criteria over the past two years. While this work has not been completed,
significant progress has been made since 2009, when a review of eligible methods was prepared (Bassler et al., 2009). The next three
sections of this paper describe these ideas, their technical justification, and sample calculations.

1 3
Lin & Yue (1990); 2 de Kat, et al. (1994); Pereira (1988); 4 Belknap & Reed (2010); 5 Spanos & Papanikolaou (2006); 6 Matusiak
(2000); 7 Bulian & Francescutto (2008); 8 Schreuder (2005); 9 Hashimoto, et al. (2011); 10 Söding (1982)

The Second Generation of Intact Stability Criteria: An Overview of Development

 Figure 1: Multi-tiered Approach for the Second Generation of Intact Stability Criteria (SLF53/WP.4 Annex 3))
VULNERABILITY CRITERIA FOR PURE LOSS OF STABILITY
Physical Background: Changing Stability in Waves
When a ship is sailing through waves, the submerged portions of the hull change. These changes may become especially significant if
the length of the wave is comparable to the length of the ship. For example, let us examine the changes that occur when the trough of a
wave is located amidships, applicable for most monohull vessels (see Figure 2a). The upper part of the bow section is usually wide,
due to bow flare, which makes the waterplane larger if the upper part of the bow section is partially submerged. The upper part of the
aft section is even larger than the bow section; thus the aft part of the waterplane becomes larger once the upper part of the aft section is
submerged. Unlike the bow and aft sections, the midship section is wall-sided for most displacement hulls. This means that very little
change occurs in the waterplane width with draft—when the wave trough is amidships, the draft at the midship section is low, but the
waterplane does not change much. As a result, when the wave trough is around the midship section, the waterplane is larger than it is
in calm water.

When the wave crest is located around the midship section, the situation changes dramatically, see Figure 2b. When the wave crest is
located amidships, a wave trough is located near the fore and aft sections. The underwater part of the bow section is usually quite
narrow, especially around the waterline, and the underwater part of the aft section is also very narrow. Thus when troughs are fore and
aft, the draft at the bow and the stern becomes shallow, which makes the waterplane very narrow.

As is well known from ship hydrostatics, the waterplane area has a significant effect on ship stability because it is directly related to the
inertia of the waterplane. If the waterplane loses area, then the GZ curve is also reduced.

Pure loss of stability is related to prolonged reduction of the GZ curve on (or near) the wave crest. This situation occurs in following
and stern-quartering seas when the ship speed is close to the wave celerity. If an additional heeling (or rolling) moment (e.g. lateral
gusty wind load, short-crested wave effect or centrifugal force due to course-keeping) is applied, a ship may attain a large roll angle and
even capsize.

The Second Generation of Intact Stability Criteria: An Overview of Development

 a)
Wave trough
Waterline with the wave amidships
trough amidships
GZ

Calm water waterline

Calm water

Calm water
b)

Calm water waterline Calm water

GZ Wave crest
amidships
Waterline with the
wave crest amidships

Figure 2: Change of Stability in Waves; a) on Wave Trough b) on Wave Crest

Level 1 Vulnerability Criteria for Pure Loss of Stability

Some hulls are more prone to pure loss of stability than others. A hull with large freeboard and significant change of geometry in the
fore and aft sections, but with a small GM value, may suffer from significant loll angle due to deterioration of stability on the wave
crest. As can be seen from Figure 3, loll angles resulted in large angles of heel when the wave crest was passing near the midship
section.

Pure loss of stability is driven by stability changes in longitudinal waves. As discussed above, certain features of the hull shape are
“responsible” for stability changes. One of the possible options proposed for the Level 1 Criterion is focused on these geometric
features (SLF 53/INF.10 Annex 5).

The criterion is the average value of the vertical wall-sidedness coefficients for the fore and aft quarter portions of the hull, both above
and below the waterline. The coefficient for vertical “wall-sidedness,” CVWS, measures the variation of the fore or aft quarter of the
waterplane area, either from the baseline to the design draft or from the design draft up to the shear line (see Figure 4). The threshold
value as the standard should be determined from sample calculations.

GZ, m Calm water GZ curve

1

0.5

0 , deg
Appearance of
significant loll angles
-0.5
0 20 40 60 80 100 120
Figure 3: Wave Passing Effect on GZ Curve of a RoPax Ship in IMO-critical Loading Conditions

The Second Generation of Intact Stability Criteria: An Overview of Development

 Figure 4: Notional Ship Profile with the Four Portions of the CVWS Considered for the Level 1 Vulnerability Assessment
The coefficient is taken relative to the maximum waterplane area of the fore or aft portion over the specified range of drafts.

d D

A WP z dz A WP z dz
(1)
below 0 above d
CCWS  ; CCWS  .
max  AWP z   d max  AWP z   ( D  d )

where d is draft, D is depth, and AWP is waterplane area of fore or after quarter of the waterplane.

The second proposed option considered is a conditional value of minimum GM calculated as follows (SLF 53/INF.10 Annex 2):
I
GM min  KB  L  KG . (2)
V
where KB is the distance of the center of buoyancy from the base line, V is the volumetric displacement calculated for the design draft,
and KG is the distance from the baseline to the center of gravity in design-loading conditions. IL is the moment of inertia of the water
plane calculated for a fraction of the draft, which normally should be determined as the intersection between the flat portion of side and
the bilge radius at amidships. Since the waterplane breadth in the midship section is almost unchanged, the waterplane area on the
wave crest can be approximated with that in calm water but with the smallest draft. As a result, the required value of IL can be taken
from an existing hydrostatic table so that no additional calculations using hull geometry are required. In this case, the threshold value
as the standard can be regarded as zero. Another advantage of the criterion is its potential ability to differentiate between multiple
loading conditions.

Both criteria are based on understanding that the vertical variability of the hull shape at bow and stern, but not at the midship section, is
responsible for stability changes in waves. The results of calculations for 40 sample ships are presented in Tables 1 and 2. The value
0.8 was considered as a tentative standard for the criterion (1) in SLF 53/INF.10 Annex 5, while GMmin>0 was proposed in the Annex 2
of SLF 53/INF.10.

Preliminary analysis needs be done while criteria (1) and (2) are still considered “draft” (as well as all other criteria considered in this
paper). First, criteria (1) and (2) seem to agree for most of the cases considered. The only exceptions are the two multi-purpose vessels
(MPVs) (L = 135 m and 105 m), the 5500 DWT bulk carrier, the 110 m tanker and both tugs.

The ONR flared topside configuration (Naval Combatant 1) represents the hull shape of a notional conventional destroyer. The
extensive operational experience of conventional destroyer hulls shows that they are not known for any pure loss of stability failures.
Therefore, both criteria are conservative in this case because the hull shape causes significant change of waterplane area, while the KG-
value is low enough to compensate for degradation of stability in a wave crest. Similar situations can be observed for the Japanese
Purse Seiner, the small tanker (L = 110 m), the 105 m MPV, and both tugs.

The situation with the 135 m MPV and the 5500 DWT bulk carrier seems to be the opposite: the change of waterplane area is not
dramatic, but the reserve KG may be insufficient to counter the decrease of stability on a wave crest.

Both criteria unanimously exclude from vulnerability large tankers and the rest of the bulk carriers; this exclusion does not require any
comment as these ship types have never suffered from pure loss of stability. Both criteria detect vulnerability to pure loss of stability
for both RoPax ships and the ONR Tumblehome topside hull form (Naval Combatant 2) for which vulnerability to pure loss of stability
seems to be plausible, as these types of ships are known for such failures (Maritime New Zealand, 2007; Hashimoto, 2009).

The Second Generation of Intact Stability Criteria: An Overview of Development

Both criteria detect vulnerability to pure loss of stability for all the tested container carriers. This seems to be result of dramatic
changes of the waterplane area for which containerships are known. It is also known that these changes are capable of causing
parametric resonance, but not necessarily pure loss of stability. The Level 2 Vulnerability Criterion, then, is expected to be tested
against this fact.

In general, it seems that criteria (1) and (2) complement each other; criterion (1) offers better accounting for the details of ship hull
shapes, while criterion (2) provides a value that is specific for a given loading condition. Choice of the appropriate set of loading
conditions remains to be done, as well as tuning computational parameters.

Table 1: Sample Calculations on Pure Loss based on Ships from SLF 53/INF.10 Annex 5
Ship Description L, m Cvwp2 GM GMmin
Bulk Carrier 1 275 0.89 6.32 4.03
Bulk Carrier 2 145 0.87 1.53 0.27
Containership 1 Post-panamax 322.6 0.73 1.30 -6.23
Containership 2 Post-panamax 376 0.74 1.84 -6.23
Containership 3 Post-panamax 330 0.73 2.08 -3.08
Containership 4 Panamax 283.2 0.73 0.46 -4.01
Containership 5 C11 Class 262 0.67 1.9 -3.01
Fishing Vessel 1 Japanese Purse Seiner 34.5 0.68 1.00 -0.84
Fishing Vessel 2 21.56 0.69 0.73 -0.19
General Cargo 1 Series 60 CB = 0.7 121.9 0.84 0.75 0.04
General Cargo 2 C4 Class 161.2 0.76 1.10 -0.53
LNG Carrier 267.8 0.85 3.40 0.34
Naval Combatant 1 ONR Flared 150 0.74 1.08 -1.75
Naval Combatant 2 ONR Tumblehome 150 0.71 2.06 -0.77
Passenger Ship 276.4 0.71 3.70 -0.49
RoPax 137 0.67 1.76 -0.58
Tanker 320 0.89 9.85 6.76

Table 2: Sample Calculations on Pure Loss based on Ships from SLF 53/INF.10 Annex 9
Ship Description L, m Cvwp2 GM GMmin
Bulk Carrier 5500 DWT 190 0.79 2.84 1.91
Bulk Carrier 180 0.8 2.10 1.17
Containership > 10000 TEU 360 0.75 0.80 -3.88
Containership > 10000 TEU 360 0.76 0.7 -3.96
Containership > 6000 TEU 320 0.74 0.70 -3.174
Containership > 6000 TEU 320 0.73 0.80 -2.88
Containership > 4000 TEU 250 0.7 0.5 -2.14
Containership > 4000 TEU 250 0.71 0.6 -2.09
Containership > 1000 TEU 210 0.71 0.60 -2.22
Containership > 1000 TEU 200 0.7 0.60 -1.98
Containership > 1000 TEU 170 0.69 0.5 -2.17
Containership > 1000 TEU 160 0.7 0.16 -1.16

The Second Generation of Intact Stability Criteria: An Overview of Development

 Table 2(Cont.): Sample Calculations on Pure Loss based on Ships from SLF 53/INF.10 Annex 9
Ship Description L, m Cvwp2 GM GMmin
Containership > 500 TEU 135 0.68 0.58 -1.00
Containership > 500 TEU 125 0.67 0.70 -1.16
Cruise Vessel 240 0.76 2.71 -1.05
LNG Carrier 1000 cbm 110 0.74 1.06 -0.11
MPV 135 0.83 0.65 -0.05
MPV 125 0.8 0.17 -0.53
MPV 120 0.71 1.00 -0.33
MPV 7500 DWT 105 0.81 0.70 -0.01
Tanker 30000 DWT 320 0.83 6.35 4.69
Tanker 110 0.72 1.31 0.6
Tug 30 0.73 2.23 0.69
Tug 25 0.67 3.60 2.01

Level 2 Vulnerability Criteria for Pure Loss of Stability

As a first approximation, pure loss of stability may be considered as a single wave event because the changes in stability are
instantaneous and do not have a memory. Typically, the worst-case wavelength is close to the length of the ship, λ/L ≈ 1.0. However,
in order to account for the effect of ship size relative to wave conditions, righting lever variations must be evaluated in irregular waves.

An irregular seaway can be presented as a series of encounters with sinusoidal waves with random length or wave number (spatial
frequency) and height or amplitude. A joint distribution of these quantities is available from Longuet-Higgins (1957, 1976, 1984),
which is based on the theory of an envelope of stochastic process (Rice, 1944/45). Based on Longuet-Higgins, each wave encounter
can be associated with a statistical weight:

ai  a k j  k
Wij    f (a, k )dkda .
ai a k j k
(3)

Here, ai and ki are amplitudes and wave numbers presented with a certain discretization over the probability density function:

a2  a2    2 2   2 2 
f ( a, k )  f ( a ) f ( k | a )  exp     exp  a (k  k1 )   exp  a (k  k1 )   .
 2V    2V (k 2  k 2 )   2V (k 2  k 2 )  
k 22  k12 2VW3  W    W 2 1   W 2 1 

where VW is the variance of the wave elevations, k1 is the mean wave number and k2 is related to the mean width of spectrum s()
expressed in terms of wave numbers using the dispersion relation:
 
1 2 1 4
k1 
VW 0
g
s() d ; k 2 
VW g
0
2
s() d .

The value of k2 includes a fourth spectral moment. Not all approximations for sea spectra allow straightforward calculation of the
fourth moment. It is known that calculations of the fourth moment for Bretschneider-type spectra are only possible if the frequency
range is limited (Bishop & Price, 1978; St. Denis, 1980).

The criterion formulated for a regular wave relates wave length and height to a measure of deterioration of stability, while this wave
passes the ship. Then the criterion for irregular waves can be sought as a mean value of a deterministic function of random variables:
wave number and wave amplitude.

The Second Generation of Intact Stability Criteria: An Overview of Development

 Ci   Cr(a , k
i j
i j )Wij .

where Cr is the criterion for a regular wave characterized with the wave number, k, and amplitude, a, while Ci is the same criterion
averaged for irregular waves defined using a given spectrum.
Alternatively, the 2nd level vulnerability check can be done using just a series of regular waves systematically covering the entire range
of possible steepness values. Here, the wavelength is assumed to be equal to the ship length as the worst-case scenario, while the range
of steepness values remains to be determined.
Three criteria for regular waves are considered below. The first criterion is based on time duration while stability is degraded due to
the passing wave.
The time while stability is decreased can be easily found if the instantaneous GM is considered a function of the wave crest position.
To evaluate this function, the instantaneous GM is calculated on a series of wave water planes corresponding to different positions of
the wave crest relative to the midship section (see Figure 5). Points x1 and x2 (Figure 6) show the distance when the GM remains
below a critical level while the wave passes the ship.

xc

Figure 5: On Calculation of the Instantaneous GM in Waves

GM change due to wave pass

x1 x2
Critical GM

Position of wave crest, m

Figure 6: Calculation of “Time-Below-Critical-GM”

The “time -below-critical GM”, tbc, can be calculated as:

x 2  x1
tbc  , (4)
c  VS

where c is wave celerity and Vs is ship speed. The time-below-critical GM is a random number in irregular waves. Its mean value is
estimated as:

m(tbc)  tbc W
i j
ij ij .

Obviously, waves that produce celerity that is too close to ship speed must be excluded to avoid singularity in Equation (4). The
criterion value Cr1 is proposed as the following ratio:

m(tbc)
Cr1  (5)
T

Here, T is a time scale of roll motions (not necessarily period in calm water) chosen to be the time scale.

The Second Generation of Intact Stability Criteria: An Overview of Development

This criterion assesses the significance of stability changes in waves. If stability is degraded only for a short duration, this degradation
may not be significant. However, for longer durations of decreased stability below the critical level, the restoring moment may be
degraded enough to result in a dangerously large roll angle. More details on this criterion as well as sample calculations on notional
naval ships is available from Belenky & Bassler (2010). This criterion may be further refined by including a simple model for surging,
as the surge motion affects phases and therefore may change the timing.
The second criterion, Cr2, is set to detect if there are significant durations of negative GM (Figure 7). Appearance of an angle of loll
may lead to the development of partial stability failure sooner, as the upright equilibrium is no longer stable. It is quite possible that
some ships may be more vulnerable to these types of failure than others.
The second criterion, is based on characteristics of time during which the angle of loll exceeds a certain limit angle, lim (30 degrees
was used in this example). For the k-th position of the wave crest along the hull crest, the indicator value, z, is calculated:

tbz   z t ,
k
k

where value, zk is an indicator, t is the time-step and index k corresponds to a particular time instant during the wave pass. For the k-
th position of the wave crest along the hull crest, the indicator value, zk, is calculated as:

0 if loll  lim
zk   .
1 if loll  lim
(30 degrees was used as lim in this example). Obviously, the angle of loll, loll, can only be obtained from the GZ curve in waves.
Calculations of the instantaneous GZ curve in waves are done in the same way as described for the instantaneous GM (Paulling, 1961).
Since the encounter frequency is low, the influence of heave and pitch can be approximated quasi-statically through balancing trim and
draft. Sometimes the GZ in waves can be approximated by using a calm-water GZ curve and the instantaneous GM in waves.
However, caution has to be exercised as there are known cases when such approximations are not conservative (Annex 9, SLF
53/INF.10).
The formulation of the second criterion is similar to the first one:
m(tbz)
Cr 2  , (6)
T

where T is the chosen time scale, and m(tbz) is the weighted average over the wave encounters:

m(tbz)   tbz W
i j
ij ij .

The third proposed criterion is based on the maximum of the GZ curve in waves. The suggested standard is zero, so the ship is
considered to be vulnerable if the GZ curve becomes completely negative at least once during the series of calculations:
Cr3  minGZ max ( xc , a, k )  0 . (7)

Sample calculations for the criteria (5), (6), and (7) can be found in Annexes 2 and 5 of SLF 53/INF.10.

GZ, m Calm water GZ curve

1

0.5

0 , deg
20 40 60 80 100

-0.5

Figure 7: Deterioration of GZ Curve near the Wave Crest

The Second Generation of Intact Stability Criteria: An Overview of Development

VULNERABILITY CRITERIA FOR PARAMETRIC ROLL
Physical Background: Parametric Resonance Caused by Changing Stability in Waves
The occurrence of parametric roll is caused by time-dependent stability changes. Typically, though not necessarily, the worst-case
condition occurs when the encounter frequency between the ship and the wave is about twice the roll natural frequency. In some cases,
the condition when the encounter frequency is close to the roll natural frequency can also be dangerous. The possibility of
development of instability due to GZ curve changes in waves has been known since the 1930’s (Watanabe, 1934, Kempf, 1938; Graff
& Heckscher, 1941; Paulling & Rosenberg, 1959).

If the ship has a non-zero roll angle while in the wave trough, increased stability provides it a strong restoring force (i.e., pushback).
As the ship returns to the upright position, its roll rate is greater since there was an additional pushback due to the increased stability. If
at that time, the ship has the wave crest at midship, the stability is decreased and the ship will roll further to the opposite side because
of the greater speed (i.e. inertia) of rolling and less resistance to heeling. Then, if the wave trough reaches the midship section when the
ship reaches its maximum roll amplitude, stability increases and the cycle starts again. An important point is that there is half the roll
cycle associated with the passing of an entire wave. So, there are two waves that pass per each roll period. That means the roll period
is about twice that of the wave period (Figure 8).

Even from this brief description, it is clear that for the development of parametric roll, two conditions are needed: a variation of
stability in waves and a certain ratio of encounter and natural frequency.

strong pushback, Stability is decreased, Stability is increased

picking up ship rolls further again, strong pushback
rotation speed cycle is repeated
20 Phase of wave

0 5 10 15 20 25 30

-20
Wave 1 Wave 2 Wave Wave 4
3
-40
Roll period 1 Roll period 2

Figure 8: Development of Parametric Roll

Level 1: Vulnerability Criteria for Parametric Roll

As the Level 1 vulnerability criteria for parametric roll is expected to be simple, it makes sense to use the Mathieu equation (the
simplest model of parametric resonance) as the basis for the criteria. The linear roll equation with periodically changing GM can be
transformed into the Mathieu equation:

I x  A44   B44  W  GM (t )  0 . (8)

The Second Generation of Intact Stability Criteria: An Overview of Development

where Ix is the transverse moment of inertia, A44 is the added mass in roll B44is the linear (or linearized) damping coefficient and W is
the weight displacement of a ship.

The variation of GM with time is the key feature to model parametric roll. In the case of regular waves, when the variation of
metacentric height is not very large, the dependence of GM on time can be approximated by a sinusoidal function:

GM (t )  GM m  GM a cos(et ) .

Here, e is the wave frequency of encounter, GMm is the mean value of the GM, and GMa is the amplitude of the GM changes in waves.
As a first approximation, GMm and GMa can be determined as:
GM a  0.5GM max  GM min  (9)

GM m  0.5 GM max  GM min  .

where GMm is the mean value of the GM, GMa is the amplitude of the GM changes in waves,and e is the wave frequency of
encounter.

The Mathieu equation is derived by transforming Equation (8) into the canonical form:

  2  2 1  h cos( t )   0 . (10)

m e

where:

W  GM m W  GM a 1 B44 2a
m  ; a  ;  ; h 2 .
I x  A44 I x  A44 2 I x  A44 m

Then the dimensionless time is introduced:

  et ;

this change of variable allows damping to be removed:

()  x()  exp  ; (11)
which finally leads to the Mathieu equation
d 2x
  p  q cos( )   x  0 , (12)
d 2
where:
 
 
p  m2   2 ; q  m2 h ;   ; m  m .
e e
The solution of the Mathieu equation (12) cannot be expressed in elementary functions. A special function, the Mathieu function, has
been introduced to describe the solution of equation (12). The most important characteristic of this solution is whether it is bounded or
not (see Figure 9).

The Second Generation of Intact Stability Criteria: An Overview of Development

 x p=0.1 q=0.2 Bounded

0.05 

0 20 40 60 80 100 120 140 160

-0.05

5 p=0.15 q=0.2 Unbounded


0 20 40 60 80 100 120 140 160
-5

Figure 9: Bounded and Unbounded Solution of Mathieu Equation

As shown in Figure 9, different combinations of the characteristics of coefficients p and q lead to either a bounded or unbounded
solution. The Ince-Srutt diagram, shown in Figure 10, maps the bounded and unbounded solutions, depending on the coefficients, p
and q. The shaded areas, identified with Roman numerals in Figure 10, correspond to the unbounded solution.

The first instability zone intersects the axis at exactly p = 0.25, which corresponds to the frequency ratio of 2; so the excitation
frequency is twice the natural roll frequency at this point. The unbounded motion belonging to this zone is commonly referred to as the
principal parametric resonance. The zoomed-in view of this zone is shown in the insert of Figure 10.

The Mathieu equation (12) has a periodic-bounded solution since damping is excluded by the substitution (11). It means that the
corresponding roll, (), decays with the damping decrement,  if x() is a periodical solution of the Mathieu equation, as shown in
Figure 9. It also means that there is a threshold value for roll damping for each pair of Mathieu parameters, p and q. If roll damping is
less than the threshold value, the roll will be unbounded as a solution of the Mathieu equation. If the roll damping is larger than the
threshold, roll is bounded, even if the solution of the Mathieu equation is unbounded. The increment of the Mathieu solution is not
enough to overcome the decrement of roll damping. It also means that with linear damping, the instability zone is narrower and requires
some finite value of GM variations even at p = 1/4; i.e., it does not touch the axis (see Figure 11).

The boundaries for indicating where parametric resonance is possible reveal two interdependent conditions that may be used to
formulate the following criteria: the frequency condition and the parametric excitation condition. The frequency condition depends on
natural and encounter frequencies. Natural frequency depends on the loading condition, while encounter frequency depends on the
wave parameters and the ship’s speed and heading. The parametric-excitation condition requires that the change of stability is large
enough to cause instability; whether this condition is satisfied depends on the hull geometry and on the parameters of the wave used for
the assessment.

q
V
III IV
6
VI
0.6 q
II
0.4 4
I
I
0.2 2

p
0 0.2 0.4
0 2 4 6 8 p

Figure 10: Ince-Strutt Diagram

The Second Generation of Intact Stability Criteria: An Overview of Development

 0.4
q
With damping 

0.2

p
-0.1 0 0.1 0.2 0.3 0.4 0.5

Figure 11: Influence of Damping on the 1st Instability Zone

To derive a criterion for the critical parametric excitation that creates roll growth, consider an approximate solution of the Mathieu-type
equation ( for exact principal resonance (e = 2m), as presented by Hayashi (1985) and used to derive a “transient stage” criterion for
parametric roll by Spyrou (2005) as follows:


(t )  e t C1e mt sinmt    C2 e  mt sinmt   .  (13)

where C1 and C2 are arbitrary constants which are determined by the initial conditions and  is a parameter controlling the growth or
decay of oscillations. It is defined as:
1 2 2
 a h  4a  12 ,
4
And a is variable expressing the frequency ratio,

2m
a4 .
e2

Obviously, a = 1 when e = 2 m, which is exactly the middle of the first instability zone (principal parametric resonance) of the Ince-
Strutt diagram.

The phase  is determined from the following expression,

2a  1 
cos 2   ;  0 .
ah 2
The encounter frequency of “dangerous” waves is considered to correspond exactly to the frequency of principal parametric resonance:
h 
a  1;    ;  . (14)
4 4
Looking at the initial condition when the initial roll rate has zero value, the constants were found equal to:

2
C1  C2   0 ;  0  0 . (15)
2
Substituting Equations (14) and (15) into Equation (13) yields:

2   h mt  
h
mt   
(t )   0 e t  e 4 sin m t    e 4 sin m t    . (16)
2   4  4 
 
Equation (16) allows calculation of the amplification factor, f, after n oscillations:

The Second Generation of Intact Stability Criteria: An Overview of Development

 1  2n 
f    . (17)
0  m 
Note that the term with the negative exponent in Equation (16) plays a small role for a motion that is in fact growing exponentially. In
this case, the substitution of Equation (16) into Equation (17) yields:

 nh 2 n   
 sin 2n   .
2
f  exp  (18)
2  2  m   4

Solving for h from Equation (18) leads to:

ln f  ln 2 4
h2  .
n m

Given an amplification factor and the number of oscillations, the following criterion is deduced, assuming the stability changes in
waves are symmetric relative to the calm water values:

GM a ln f  ln 2 4
2  . (19)
GM 0 n 0

The parameters  and 0 in Equation (19) have to be chosen. If no other data are available, ABS (2004) recommends, as a conservative
estimate,

 0.03 . (20)
0

The number of cycles and the amplification factor are obviously related. Larger amplification of the initial roll may be expected for
more cycles. These parameters are very important for fine-tuning the criteria and need to be addressed during a later stage of criterion
development. As a preliminary guess of f = 5 and n = 4 leads to,
GM a
 0.5 . (21)
GM 0

Equation (19) can also be used to derive a criterion without transient effects. Such a criterion identifies the existence of an unbounded
solution, as the solution tends to go to infinity with an increase in the number of cycles,
 ln f  ln 2 4  4 GM a 4
lim  2     . (22)
n  n 0  0 GM 0 0
f 

Using a conservative assumption on roll damping in Equation (20) leads to

GM a
 0.12 . (23)
GM 0

As can be expected the exclusion of transient effects leads to a far more conservative standard. The importance of Equation (23) is that
it sets a practical limit for the “conservativeness” of the standard, as the damping assumption in Equation (20) is extremely
conservative. Another important point is that including transient effects in the criteria is similar to assuming less conservative
damping. This effect is presented in detail in Appendix 1.

These criteria require knowledge of the magnitude of parametric excitation in Equation (9), which is based on calculations of the
instantaneous GM in waves. While these calculations are straightforward, they involve computer software and additional time needed
for preparation of geometric input. As was discussed above, most of the changes of the waterline for ships come from the bow and
stern quarters of the hull, and these changes may be revealed by just altering the draft. Altering the draft was used for the Level 1
Vulnerability Criterion for pure loss of stability (2) proposed in SLF53/INF.10-Annex 2:

I up  I low
GM a  , (24)
2V
where Iup and Ilow indicate the moments of inertia of the water plane for upper and lower drafts, respectively, and V is the displaced
volume of the ship for a standard draft. It may be assumed the upper draft is lower than the freeboard’s deck by 5% of freeboard, and

The Second Generation of Intact Stability Criteria: An Overview of Development

the lower draft is half the mean draft or the intersection between the bilge circle and the side hull wall midship. Other proposals for
these reference drafts can be found in SLF 53/3/9.

An obvious advantage of the criterion expressed in Equations (21) and (24) is its simplicity. The criterion can be evaluated without
extra calculations, as all the values can be directly obtained from the vessel’s hydrostatic curves. At the same time, this criterion
accounts for neither the ship’s forward speed nor the likelihood of encountering dangerous environmental conditions. These factors
may be determined by using the frequency of encounter for condition of parametric roll and by specifying the waves for the reference
calculation.

The frequency of encounter can be determined using the boundaries of the first instability zone of the Ince-Strutt diagram. Because
oscillations grow inside the instability zone and decay outside the instability zone, the solution has to be periodic at the boundary. To
account for the damping already included in Equation (13), it is enough to satisfy:

h
 m ,
4
to obtain the well-known approximation for the boundary of the first instability zone of Ince-Strutt diagram:
2 2
 1 e2   e  
h  2 1     . (25)
 4 2   2 
 m   m 
This approximation establishes the relationship between coefficient q that is related to the magnitude of parametric excitation (9) and
coefficient p, the ratio of the natural frequency to the frequency of encounter,

e   
2
g
Vm cos  
2

 g  V m 
2 cos  ,

where Vm is the forward speed in m/s,  is heading angle relative to the waves (0 is following seas), g is the acceleration due to gravity,
 is the true frequency of the wave, and  is the wave length. Since the frequency of encounter depends on speed, Equation (25) can
be used to determine if the design speed of a ship falls into a frequency range where parametric roll is possible,
g  2  2  h 
Vm1,m 2   , (26)
2 T

where T is the roll period.

A more sophisticated criterion can be proposed if actual wave parameters are taken into account. To account for the likelihood of
encountering different waves, a series of wave cases could be used, as proposed in Annex 1 SLF 53/INF.10 (see also Bulian &
Francescutto, 2010). The final form of the criterion is similar to:

C PR1  W C(H ,  ) .
i
i i i (27)

Each wave case is defined by wave height Hi, and length, i and Wi is the statistical weight of a given wave case. For each wave case
the criterion is
 GM a ( H i ,  i )
1, if  r and V PR  VD
Ci   GM m ( H i ,  i ) ,
 (28)
0, otherwise,

where
2 g 
VPR  
T 2

and r is the ratio between the magnitude of GM changes and the mean value of magnitude providing the specified f-fold increase after n
roll cycles (preliminary value is 0.5); VD is the design speed. VPR is the speed where parametric roll is expected (wave encounter
frequency is twice the roll natural frequency). Cases with negative GMm require special handling.

The Second Generation of Intact Stability Criteria: An Overview of Development

The application of the criterion for parametric roll in Equation (27) obviously requires more computational work than evaluating
Equation (21). However, these efforts may very well pay off if employing the application of Equation (27) indicates an absence of
vulnerability to parametric roll, when the criterion in Equation (21) may be too conservative.

Sample calculations for the parametric roll Level 1 Vulnerability Criteria are presented in Tables 3 and 4. Following the suggestion
from SLF 53/3/7, the speed range criterion in Equation (26) was applied to the sample population of ships from SLF 53/INF.10 Annex
5, while a very conservative assumption was used for the magnitude of parametric excitation (h = 1) (see Table 3). As a result of using
this conservative assumption, only the tanker is found not vulnerable.

The criterion and standard in Equation (21) are carried out using a wave with a length equal to ship length and with the steepness taken
from SLF 53/3/7. The resolution of this criterion seems to be better, as it excluded bulk and LNG carriers and both fishing vessels in
addition to the tanker; these ships do not have a known vulnerability to parametric roll (SLF 53/INF.10 Annex 5). Application of the
approximation given in Equation (24) for the magnitude of the parametric excitation provides results that are similar in terms of
resolution. Sample calculations for the criterion in Equation (28) are available from Bulian & Francescutto (2011).

Table 3: Sample Calculations on Parametric Roll, Level 1, based on Ships from SLF 53/INF.10 Annex 5
Criterion (26), h=1 Criterion and Standard (21)
GM, Vs, GM in waves Eq (24)
Ship L, m m kn Vs1, kn Vs2, kn Y/N Value Y/N Value Y/N
Bulk Carrier 275 3.46 15 2.90 -24.48 Y 0.09 N 0.36 N
Bulk Carrier 2 145 0.56 15 12.96 1.02 Y 0.40 N 1.43 Y
Containership 1 322.6 1.27 25 16.22 -3.85 Y 1.42 Y 4.56 Y
Containership 2 376 1.79 25 17.05 -4.95 Y 0.90 Y 4.44 Y
Containership 3 330 1.85 25 10.31 -14.45 Y 0.58 Y 2.40 Y
Containership 4 283.2 1.15 25 8.43 -15.33 Y 0.75 Y 2.60 Y
Containership 5 262 2.00 25 7.45 -15.89 Y 0.77 Y 3.32 Y
Fishing Vessel 1 34.5 1.70 18 -6.08 -20.98 Y 0.16 N 0.47 N
Fishing Vessel 2 21.56 0.80 15 1.74 -5.24 Y 0.09 N 0.77 Y
General Cargo 1 121.9 0.67 18 14.40 5.29 Y 1.03 Y 1.03 Y
General Cargo 2 161.2 0.90 16 7.85 -8.99 Y 0.57 Y 1.54 Y
LNG Carrier 267.8 3.42 18 -1.16 -31.11 Y 0.18 N 0.62 Y
Naval Combatant 1 150 1.04 30 1.73 -18.79 Y 0.58 Y 5.25 Y
Naval Combatant 2 150 2.06 30 -18.0 -52.94 Y 0.60 Y 0.89 Y
Passenger Ship 276.4 3.75 25 -16.75 -58.59 Y 0.57 Y 0.97 Y
RoPax 137 1.79 18 -2.65 -25.41 Y 0.82 Y 1.40 Y
Tanker 320 9.95 14 -16.36 -60.16 N 0.03 N 0.21 N

Table 4 contains the results of calculations for the sample ship population from SLF 53/INF.10 Annex 10. The results in Table 4
(shown for two loading conditions) are qualitatively similar to those in SLF 53/INF.10 Annex 10. All the container carriers and the
cruise vessel are found to be susceptible to parametric roll, while tugs, tankers and bulk carriers are not. Generally, this is consistent
with existing operational experience. Multi-purpose vessels and the LNG carrier show vulnerability in certain loading conditions; thus
their vulnerability must be determined by the level 2 criteria. Results for more loading cases are available for these ships in
SLF 53/INF.10 Annex 10.

Level 2 Criteria for Parametric Roll

The linear equation of roll motion, Equation (8), as well as its particular case in Equation (10), is a good model to indicate the onset of
parametric roll. However, it is not capable of estimating the amplitude of parametric roll; once parametric resonance in a linear system
has started, its amplitude grows without any limit. Nonlinearity stabilizes parametric resonance at certain amplitude (see Figure 12).

The Second Generation of Intact Stability Criteria: An Overview of Development

The mechanism of stabilization is related mostly to the nonlinearity of the GZ curve. Once the amplitude of the roll oscillation
becomes large enough, the frequency condition for parametric resonance is no longer satisfied, as the change in instantaneous GM
leads to a change in the instantaneous natural frequency. (It is a well-known phenomenon of nonlinear roll—the dependence of natural
frequency on amplitude is also known as the loss of isochronous properties.) This means that once a certain roll amplitude has been
exceeded, further flow of energy into the dynamical system stops. Additional energy is available only below this amplitude, which
leads to the establishment of the energy/work balance and stabilization of the amplitude, and eventually leads to a limit cycle.
Therefore, the simplest mathematical model capable of reproducing stabilization of parametric roll must include the nonlinear GZ
curve:

I x  A44   B44  W  GZ (, t )  0 ,

where GZ(,t) means a full GZ curve changing while the wave passes (see Figure 3).
Nonlinear roll damping can also be a factor of stabilization. Its mechanism is slightly different from one with the GZ curve. An
increase in the amplitude of roll motions leads to an increase in the roll rate magnitude when the ship crosses the equilibrium point. The
increase in roll rate makes nonlinear damping large enough to elevate the threshold above the current level of parametric excitation.
This also stops the flow of energy to the dynamical system and stabilizes the parametric roll.

Table 4: Sample Calculations on Parametric Roll, Level 1, based on Ships from SLF 53/INF.10 Annex 10
Ship Description L, m d, m GM1 Eq (24) Y/N GM2 Eq (24) Y/N

Bulk Carrier 5500 DWT 190 12.8 2.84 0.2 N 3.34 0.17 N
Bulk Carrier 180 0.75 2.10 0.33 N 2.60 0.26 N
Containership > 10000 TEU 360 16.0 0.80 3.76 Y 1.30 2.31 Y
Containership > 10000 TEU 360 15.5 0.70 3.87 Y 1.20 2.26 Y
Containership > 6000 TEU 320 14.6 0.70 4.28 Y 1.20 2.5 Y
Containership > 6000 TEU 320 14.5 0.8 3.18 Y 1.30 1.96 Y
Containership > 4000 TEU 250 13.5 0.50 3.03 Y 1.00 1.52 Y
Containership > 4000 TEU 250 12.5 0.60 2.96 Y 1.10 1.62 Y
Containership > 1000 TEU 210 12.0 0.60 2.71 Y 1.10 1.48 Y
Containership > 1000 TEU 200 11.4 0.60 3.19 Y 1.10 1.74 Y
Containership > 1000 TEU 170 10.9 0.50 3.6 Y 1.00 1.8 Y
Containership > 1000 TEU 160 10.5 0.16 7.88 Y 0.66 1.91 Y
Containership > 500 TEU 135 7.75 0.58 1.77 Y 1.08 0.95 Y
Containership > 500 TEU 125 7.40 0.70 1.96 Y 1.20 1.14 Y
Cruise Vessel 240 7.20 2.71 0.86 Y 3.21 0.73 Y
LNG Carrier 1000 cbm 110 8.48 1.06 0.69 Y 1.56 0.47 N
MPV 135 8.00 0.65 0.91 Y 1.15 0.51 Y
MPV 125 7.00 0.17 3.23 Y 0.67 0.82 Y
MPV 120 7.08 1.00 0.98 Y 1.50 0.65 Y
MPV 7500 DWT 105 7.06 0.7 0.64 Y 1.20 0.37 N
Tanker 30000 DWT 320 22.6 6.35 0.08 N 6.85 0.08 N
Tanker 110 7.41 1.31 0.32 N 1.81 0.23 N
Tug 30 3.50 2.23 0.4 N 2.73 0.33 N
Tug 25 2.43 3.6 0.2 N 4.10 0.18 N

The Second Generation of Intact Stability Criteria: An Overview of Development

 60 , deg Nonlinear equation with periodically
Mathieu (linear) equation changed GZ curve
40

20
t, s
0 20 40 60 80 100 120 140
-20

-40

-60

Figure 12: Modeling of Parametric Roll with Mathieu Equation (Blue) and Nonlinear Equation (Red)
However, roll-damping nonlinearity is relatively weak compared to GZ curve effects and should be considered as a secondary factor for
stabilization in regular waves. (In irregular waves, the contribution from nonlinear roll damping may be different.) Also, if the roll
amplitude is large enough, damping becomes dependent on the roll angle. This dependence may lead to a decrease of damping (when a
bilge keel emerges from the water), as well as to an increase in damping (when the deck edge submerges into the water). Hence, this is
dependent on ship-specific geometry (Bassler, et al., 2011). These considerations, however, make the mathematical model too
complex for the second level vulnerability check, so it makes sense to limit the description of damping to a cubic or quadratic
approximation:

  f ( )  2  f (, t )  0 , (29)

d 

where
f d ( )  2   3 3

or
f d ( )  2  sign( ) 2  2 .

2 and 3 are the linear, quadratic and cubic damping coefficients, respectively; and f(,t) is a time-dependent stiffness term related
to the instantaneous GZ curve in waves. The damping coefficients can be found either from a roll decay test or by methods like Ikeda’s
semi-empirical formula.

As mentioned above, the calculation of the instantaneous GZ curve in regular waves is straightforward. However, the assumption of
low-encounter frequency is not always applicable for parametric roll as it is for pure loss of stability, since the former may occur in
head and bow-quartering seas as well as in following and stern-quartering seas.

The attitude of a ship is calculated based on the heave and pitch response to a regular wave:

M  A33  G  B33 G  A35  B35  F ( G , , t )  FH (t )

    
 (30)
I Y  A55   B55  A53 G  B53 G  M  ( G , , t )  M H (t ) ,
    

where M is mass of the ship; IY is the mass moment of inertia relative to the transverse axes; A33 and A55 are heave-added mass and
pitch moment of inertia (assumed to be equal to the corresponding mass and moment of inertia), respectively; and B33 and B55 are
damping coefficients for heave and pitch. FH and MH are hydrodynamic components of the wave excitation. Functions F and M are
the difference between hydrostatic and Froude-Krylov forces and moments, respectively, at the instant of time, t. These values are
expressed as follows:
 0.5 L 
F ( G , , t )  g V0  ( x, z ( G , , t )) dx 


 
 0.5 L 
and

The Second Generation of Intact Stability Criteria: An Overview of Development

  0.5 L 
M  ( G , , t )  g V0  LCB0  M  ( x, z ( G , , t )) dx  ,

 
 0.5 L 
where  is mass density of water, V0 is volumetric displacement in calm water, and LCB0 is the longitudinal position of the center of
buoyancy in calm water. Functions  and M calculate an area and a static moment relative to the y-axis of a station located at a
longitudinal position along the hull, x. The second argument of this function shows the submergence of this position along the hull, as
expressed by the function of the instantaneous waterline z(G,,t) (see Figure 13).

z(G,,t)

Figure 13: Sample Instantaneous Waterlines Evaluated from Heave and Pitch Response
Once the Froude-Krylov terms are defined, the system of differential Equations (30) can be integrated with a standard Runge-Kutta
solver. Once the steady state of the response is reached, the instantaneous attitude of the ship for each position of the wave crest along
the hull can be determined.

While this method for evaluating the heave and pitch attitude for stability in waves seems to be the most physically sound, several
assumptions must be made regarding the heave and pitch hydrodynamic coefficients. If a seakeeping analysis has been performed at
this stage in the design process, the added mass and damping are already available. If the vulnerability to parametric roll needs to be
addressed earlier in the design process, additional assumptions are inevitable. The heave-added mass and pitch-added moment of
inertia can be assumed equal to the corresponding mass and moment of inertia of the vessel. Simple assumptions (subject to further
verification) can also be made for heave and pitch damping, based on typical values as a percentage of critical damping (40–50%).
Finally, the diffraction excitation and hydrodynamic coupling coefficients can be neglected at this stage. It is quite simple to test these
assumptions by direct comparison between the results of the simplified calculations and complete potential flow solutions. Such testing
remains on the agenda for the future work.

Once the attitude of the ship for each wave crest position has been determined, the GZ curve in waves can be evaluated. To calculate
the GZ curve in waves, the current attitude of the vessel on the wave is considered to be in equilibrium following d’Alembert’s
Principle, where the dynamic problem can be considered as a static problem if the inertial forces are added. These inertial forces show
themselves as the difference between the current attitude and the equilibrium attitude on the wave for the same position of the wave
crest. Further calculations of the GZ curve in waves are done in a usual way, utilizing the balancing of trim and draft at each heel
angle. Models accounting for dynamic heave have also been developed and used by Kroeger (1986), Jensen (2007), and others.

Once the GZ curve in waves is available, roll Equation (29) is solved. It is a nonlinear ordinary differential equation and does not have
a known exact solution, leaving two options: an approximate analytical or a direct numerical solution. Both options allow
consideration of nonlinear damping along with nonlinear restoring.

Approximate analytical methods for solution of the roll equation have been used extensively in the past (e.g. Sanchez & Nayfeh, 1990;
Oh, et al., 2000; Bulian, 2004; Spyrou, 2005) and have proven themselves to be quite efficient for relatively simple models. For
example, the restoring function is considered in the following form:

 2
f (, t )    l33  l55 
1
GM m  GM  GM a cos e t    1      ,
GM   
where l3 and l5 are polynomial coefficients for the GZ curve in calm water. An approximate solution based on the first harmonic can be
found using a well-known method such as “averaging” or “harmonic balance”, as shown by:

  a cos t   ;   e / 2 .

Then, the following algebraic equation can be obtained for deriving the steady-state amplitude, a, of the periodic solution.

The Second Generation of Intact Stability Criteria: An Overview of Development

 2
1  6 a   8  (GM m  GM )
2 2 2 2
 3 2 2   8 
 4 2  a 2 3 2    
 2    16   2
  2
GM
 a   a
(31)
2

5 4a l5 2  6 2a l32  8 2  2 


2
4  GM a 
 
2
.

 2   2a   GM 

It can be shown that Equation (31) yields a linear solution if:
l3  l5  GM m  0 ;   0 . (32)
Substitution of the above expressions in Equation (32) into Equation (31) reduces the latter to:
8 2  E GM a
2
 , (33)
( 2   2a )  0 GM

where E is the equivalent linear damping. Equation (33) is consistent with the level 1 criterion in Equation (22). This can be shown by
letting the amplitude, a, be small. Then, the equivalent linear damping, E, becomes linear damping,  and with the square of the roll
amplitude being small relative to 22, Equation (33) reduces to Equation (22).

Numerical solution of Equation (29) can directly utilize the results of the calculation of the GZ curve in waves. Another advantage of
the numerical approach is the possibility of obtaining a transient solution, which avoids unnecessary conservatism in cases where
parametric roll grows so slowly that the ship does not reach a large-roll angle in irregular seas (cf. Peters, et al., 2010; also Annex 5 of
SLF 53/INF.10). Application of a wave group, consisting of a limited number of waves with the same length and height, leads to the
same result as the “typical” wave group. Using such wave groups limits the calculation of stability in waves to only one type of wave.
This significantly reduces the otherwise substantial number of calculations needed in the numerical approach (Annex 9 of SLF
53/INF.10).

Sample calculations using a direct numerical solution are shown in Tables 5 and 6. These calculations are based on the sample ship
population from Annexes 5 and 10 of SLF 53/INF.10, respectively. These results are of a preliminary nature as the procedure for
calculations has not yet been established. The initial heel angle used in Table 5 is 10 degrees, while 5 degrees is used in Table 6. The
initial roll rate is zero in both cases. Sample calculations using the approximate analytical solution in Equation (31) can be found in
Annex 2 of SLF 53/INF.10.

The main result of these calculations is an observed maximum roll angle. In a case of no parametric resonance, roll motions will decay
and the maximum roll angle will be the initial angle. This will be the case for tankers, bulk and gas carriers, and naval combatants.
Quick development of parametric roll is observed for container carriers, RoPax and cruise vessels; and these results are consistent with
existing experience.

General cargo and multi-purpose vessels as well as fishing vessels are in a “grey” area. While the two sample fishing vessels in the
above table do not indicate any vulnerability to parametric roll, fishing vessels, in principle, are known to be susceptible to parametric
roll (e.g. Neves, et al., 2009). In principle, general cargo ships can exhibit parametric roll in following seas under the condition of
insufficient stability (Paulling, et al., 1972, 1974, 1975). Some differences in the severity of the parametric roll response are observed
in the case of MPVs, depending on their loading conditions (see Table 6).

The question remains about what roll angle should be used as a standard. The ABS Guide (2004) used 22.5 degrees; this value came
from a limit on the main engine operability for container carriers. It is based on lubrication requirements at static inclinations. Another
standard might be proposed based on container-lashing requirements; this seems to be logical, but it may not be an appropriate criteria
basis for other types of vessels. While the issue of the allowable roll angle needs further discussion, 20 degrees is used here to indicate
vulnerability to parametric resonance.

An important observation is that the result of sample calculations for level 2 vulnerability criteria is, in general, less conservative than
the level 1 vulnerability criteria. The results are consistent because all of the ships found to be non-vulnerable by the level 1 criteria
have been found non-vulnerable by the level 2 as well.

The Second Generation of Intact Stability Criteria: An Overview of Development

 Table 5: Sample Calculations on Parametric Roll, Level 2, based on Ships from SLF 53/INF.10 Annex 5
Ship Description L, m GM, m  Roll Y/N
Bulk Carrier 275 9.4 0.05 10 N
Bulk Carrier 2 145 0.53 0.05 10 N
Containership 1 Post-panamax 322.6 1.11 0.05 >25 Y
Containership 2 Post-panamax 376 1.84 0.05 >25 Y
Containership 3 Post-panamax 330 1.64 0.05 >25 Y
Containership 4 Panamax 283.2 1.06 0.05 >25 Y
Containership 5 C11 Class 262 1.91 0.05 >25 Y
Fishing Vessel 1 Japanese Purse Seiner 34.5 1.69 0.05 10 N
Fishing Vessel 2 21.56 0.73 0.05 10 N
General Cargo 1 Series 60 CB = 0.7 121.9 0.24 0.05 10 N
General Cargo 2 C4 Class 161.2 1.10 0.05 14.9 N
LNG Carrier 267.8 3.40 0.05 10 N
Naval Combatant 1 ONR Tumblehome 150 1.03 0.15 10 N
Naval Combatant 2 ONR Flared 150 3.01 0.15 10 N
Passenger Ship 276.4 3.70 0.05 >25 Y
RoPax 137 1.77 0.05 >25 Y
Tanker 320 9.76 0.05 10 10

Table 6: Sample Calculations on Parametric Roll, Level 2, based on Ships from SLF 53/INF.10 Annex 10
Ship Description L, m d, m GM1 Roll Y/N GM2 Roll Y/N
Bulk Carrier 5500 DWT 190 12.8 2.84 5 N 3.34 5 N
Bulk Carrier 180 0.75 2.10 5 N 2.60 5 N
Containership > 10000 TEU 360 16.0 0.80 32 Y 1.30 32 Y
Containership > 10000 TEU 360 15.5 0.70 >35 Y 1.20 >35 Y
Containership > 6000 TEU 320 14.6 0.70 32 Y 1.20 32 Y
Containership > 6000 TEU 320 14.5 0.8 32 Y 1.30 32 Y
Containership > 4000 TEU 250 13.5 0.50 31 Y 1.00 31 Y
Containership > 4000 TEU 250 12.5 0.60 32 Y 1.10 32 Y
Containership > 1000 TEU 210 12.0 0.60 32 Y 1.10 31 Y
Containership > 1000 TEU 200 11.4 0.60 32 Y 1.10 31 Y
Containership > 1000 TEU 170 10.9 0.50 32 Y 1.00 32 Y
Containership > 1000 TEU 160 10.5 0.16 32 Y 0.66 30 Y
Containership > 500 TEU 135 7.75 0.58 32 Y 1.08 31 Y
Containership > 500 TEU 125 7.40 0.70 31 Y 1.20 30 Y
Cruise Vessel 240 7.20 2.71 31 N 3.21 30 N
LNG Carrier 1000 cbm 110 8.48 1.06 10 Y 1.56 5 N
MPV 135 8.00 0.65 23 Y 1.15 9 N
MPV 125 7.00 0.17 30 Y 0.67 13 N
MPV 120 7.08 1.00 30 Y 1.50 22 Y
MPV 7500 DWT 105 7.06 0.7 16 N 1.20 7 N
Tanker 30000 DWT 320 22.6 6.35 5 N 6.85 5 N
Tanker 110 7.41 1.31 5 N 1.81 5 N

The Second Generation of Intact Stability Criteria: An Overview of Development

VULNERABILITY CRITERIA FOR SURF-RIDING AND BROACHING-TO
Physical Background of Surf-riding and Broaching-To
Broaching-to is a violent uncontrollable turn that occurs despite maximum steering effort. As with any other sharp-turn event,
broaching-to is accompanied by a large heel angle, which may lead to partial or total stability failure. Broaching-to is mostly
associated with operating in following and stern-quartering seas. Stability failure caused by broaching-to is known as a problem
mainly for fishing vessels and high-speed, monohull passenger ships.

Broaching-to is often preceded by surf-riding. Surf-riding occurs when a wave, approaching from the stern, “captures” a ship and
accelerates the ship to the wave-phase speed (wave celerity). To the outside observer, surf-riding looks like a transition from periodic
surging (when waves overtake a ship) to a situation where a ship runs with a wave.

This transition is a well-known and established nonlinear phenomenon that has been discussed several times in the literature (e.g. Grim,
1951; Kan, 1990). From a nonlinear dynamics perspective, the fundamental dynamic sequence that leads to surf-riding and then to
broaching-to has been identified by Spyrou (1996, 1997). As this phenomenon is well understood, at least in a deterministic
environment, it seems viable to develop meaningful and scientifically sound criteria expressing the likelihood of surf-riding.

As surf-riding is in most cases a stationary condition, the wave profile does not vary relative to the ship. Furthermore, during surf-
riding some ships exhibit dynamic yaw instability, despite active control. This leads to the uncontrollable turn identified as broaching-
to. Therefore, the likelihood of surf-riding could be used to formulate suitable vulnerability criteria for broaching-to. For surf-riding to
occur, the wave length is usually comparable with the ship length. Large ships cannot surf-ride because waves of the necessary length
are simply too fast compared to the ship speed.

Surf-riding is an equilibrium condition in which the sum of the wave-induced surge force, propeller thrust (at given speed settings) and
resistance is zero, with the ship speed equal to the wave celerity. As can be seen from Figure 14, there are two crossings and, therefore,
two equilibria on the span of one wave. One of these equilibria is located near the wave crest. It is unstable in surge in that if a ship is
perturbed from this position, it will continue the motion rather than returning to the equilibrium position. The other equilibrium
position is located near the wave trough and is stable in surge, in that if a ship is perturbed from this position, it will return to the
equilibrium position.

There are two characteristic speed settings, or nominal Froude numbers, associated with surf-riding. One is called the critical speed for
surf-riding under certain initial conditions (or the first critical speed, or the first threshold), and the other is critical speed for surf-riding
under any initial conditions (Makov, 1969). The critical speed for surf-riding under certain initial conditions is where surf-riding
equilibrium exists.

Surging motions, however, are still possible, even when surf-riding equilibria exist, as shown in the phase plane diagram (Figure 15).
Some combinations of the ship’s position on the wave and its instantaneous forward velocity correspond to surf-riding while others
combinations correspond to surging.

The origin of the coordinate system in Figure 15 is located on a wave crest and therefore, it moves with the wave celerity. Surf-riding
corresponds to the points on the x-axis relative to the position of the wave crest (at which the ship is moving at the wave celerity—G is
zero). While the ship is surging, its mean speed is less than the wave celerity (i.e., G is negative), which means that the surging
trajectories are shown as dashed lines leading backwards and away from the wave crest. The shaded area in Figure 15 shows
combinations of a position on the wave and instantaneous forward speed relative to the wave celerity that lead to surf-riding; the rest of
the phase plane corresponds to surging.

Surging and surf-riding do not coexist for all speeds. Increasing the ship’s speed causes the surf-riding area to grow, until the surging
area ceases to exist (see Figure 16). The speed (or nominal Froude number) corresponding to that situation is regarded as the critical
speed for surf-riding under any initial conditions, or the second critical speed (or the second threshold). If one considers the phase
plane comprised by the pair of ship’s longitudinal position and velocity, surf-riding will appear as occurring under any initial
conditions (somewhere between 23.25 and 23.5 knots in Figure 16).

For a general discussion, these critical speeds (the first one appears when surf-riding becomes possible; and the second (higher) one,
when surf-riding becomes inevitable) are more conveniently expressed in terms of the nominal Froude number rather than as
dimensional speeds. These two Froude numbers are the major characteristics of the likelihood of surf-riding. Wave capture associated
with surf-riding is not just an acceleration phenomenon; in fact, if wave capture occurs while going faster than the wave celerity,
deceleration will occur. However, this type of surf-riding is only relevant for very high-speed craft, so it is not discussed here.

The Second Generation of Intact Stability Criteria: An Overview of Development

 2000 F, kN
Positions on the wave

1000

Wave Phase

0 100 200 300 G, m

Difference between
thrust and resistance
Unstable
Equilibrium Stable
Wave surge force
Equilibrium
-2000

Figure 14: Surf-Riding Equilibria for a 100 m High-speed Vessel, Wave Height 6 m, Wave length 200 m, Speed Setting 24 kn

20 
G, m/s

10
Surf-riding
Surf-riding

300 -100 0 100 G, m

Surging Surging
10

Figure 15: Phase Plane with Surging and Surf-Riding (Belenky, et al., 2008)

a) VS=18 kn  b) VS=21 kn

G G

G G

c) VS=23.25 kn  
G c) VS=23.50 kn G

G G

Figure 16: Changes of the Phase Plane with Increase of Speed Settings

The Second Generation of Intact Stability Criteria: An Overview of Development

Mathematical Model for Surf-riding
Because surf-riding is often a prerequisite to broaching-to, the development of the vulnerability criteria focuses on surf-riding, as the
mathematical modeling is simpler than for broaching-to.

Identification of the first critical speed is simple, as the calculation of the Froude-Krylov surge-wave force is straight forward. This
force is a result of the integration of the incident wave pressure in the absence of the ship over the surface of the hull (a derivation is
shown in Belenky & Sevastianov, 2007):

FW (G )  gk A[ AS sin(kG )  AC cos( kG )] (34)

where
0.5 L 0.5 L
AS  
0.5 L
A0 ( x)  cos( kx )dx ; AC   A ( x)  sin(kx)dx ,
0.5 L
0 (35)

0
A0 ( x)  2  y( x, z) exp( kz )dz ,
d ( x)
(36)

and x, y and z are the coordinates of points on the surface of the hull, expressed in a ship-fixed coordinate system; y(x, z) is the half-
breadth on a station with coordinate x at depth z; d(x) is the draft of a station at longitudinal position x; k is the wave number; A is the
wave amplitude; and  is the mass density of water.
The above prediction method shows reasonable agreement with model experiments except for ships with very fine forms where it is
believed that nonlinear wave effects may be significant. While effects of nonlinear incident waves are rather limited, even in steeper
waves; three-dimensional diffraction effects may be important for some ships (Sadat-Hosseini, et al., 2011).
Once the amplitude for the wave-surging force is calculated, the first threshold can be found by solving the following algebraic
equation relative to speed (thrust) setting, expressed as the commanded rps, n.

R(c)  T (c, n)  gk A AS2  AC2  0 . (37)

where R is the resistance in calm water, T is the, and c is the wave celerity.
While the calculation of the first threshold is simple, it may not be the most appropriate choice for setting a criterion because, when the
equilibrium first appears, its domain of attraction is small (see Figure 16a). Also, the first threshold is higher than the surging trajectory
for the nominal speed and, as a result, the ship is not likely to enter this domain without additional force being applied. Thus, a
criterion derived from Equation (37) would unnecessarily penalize ships.
Calculation of the second threshold is more involved, as it requires examining the behavior of the boundary between the domains of
surf-riding and surging. To understand this, the simplest mathematical model describing surf-riding and surging can be considered:

(m  mx )  G  R(c   G )  T (c   G , n)  FW (G )  0 . (38)

where m is the mass of the ship, mx is the surge added mass, G is the distance between the wave crest and the center of gravity of the
vessel, c is the wave celerity, and n the commanded rps of the propeller (i.e., the speed or thrust setting) is an independent parameter.
Equation (38) can be solved numerically for a given pair of initial conditions: ship position on the wave and instantaneous speed (i.e., a
single point in the phase plane). The equation needs to be solved only long enough to indicate whether the ship is surging or surf-
riding. The initial conditions can be plotted on a graph: e.g. black for surf-riding; white for surging. Placing these initial conditions in
the nodes of a grid, it is possible to find an approximate location of the boundary. The process must be repeated for different speed
settings until all the initial conditions produce surf-riding. Since determining the threshold is the objective, the calculations can be
limited to few indicative pairs of initial conditions with relatively low-initial instantaneous speed.
Another way to determine the threshold is by detecting the change of the shape of the boundary between the domains of surf-riding and
surging. This approach requires determining the boundary; comprised of the unique trajectory, that leads towards unstable equilibrium.
Positions of equilibria are found from Equation (38) by letting G   G  0 ,

R(c)  T (c, n)  FW (G )  0 . (39)

The Second Generation of Intact Stability Criteria: An Overview of Development

If the speed settings are above the first threshold, Equation (39) must have two solutions (GS; GU), corresponding to stable and
unstable equilibria, respectively (see also Figure 14). The boundary between surging and surf-riding can be found by integration
starting from an unstable equilibrium. Since the difference between the wave and ship speed at unstable equilibrium is zero, the initial
conditions for integration need to be chosen, slightly shifted from unstable equilibrium, in the Eigen-direction defined by the following
equation that refers to the local-linearized system (see Figure 17):


x    x   2x  kWL  x . 
The coefficients x and kWL are determined by local linearization at the unstable equilibrium, GU,

1 dFW ( G ) 1 dR( G )
kWL   , x 
(m  m x ) d G  G   GU 2(m  m x ) d G  G  0

The application of Newton’s method allows the direct detection of this threshold (Umeda, et al., 2007).

An analytical, approximate but highly nonlinear method, the so-called Melnikov’s method, can produce a practical criterion of surf-
riding in closed form (Kan, 1990; Spyrou, 2006). The idea is based on the fact that the boundaries overlap when the ship speed (i.e.,
thrust) settings correspond to the second threshold. Melnikov’s function reflects the distance in the phase plane between the two
boundaries. Therefore, the instant of achieving the second threshold corresponds to the zero-value of Melnikov’s function.

For a dynamical system, a closed-form expression of the Melnikov function is sometimes approximated as a perturbation from a
Hamiltonian system. Practically, this means that the system should be relatively lightly damped, while a Hamiltonian system can
include nonlinearity in the stiffness term. In contrast with other perturbation methods, Melnikov’s method does not require small
nonlinearity in the restoring term in order to be applicable. One can also relax the requirement of low damping, but this would be at the
expense of not obtaining a closed form solution. To apply Melnikov’s method, thrust and resistance are expressed with elementary
functions and Equation (38) is transformed into the following non-dimensional form:

x  p1 (n) x  p2 x 2  p3 x3  sin x  r (n) ; x  kG .

where x  kG , with k the wave number; r the quasi-static part of the balance between thrust and resistance; coefficients, p1, p2, and p3,
represent the change of this balance; and n is the current speed (thrust) setting.

As shown in (Spyrou, 2006), Melnikov’s function can be expressed as,

4 32
M (n)  r (n)  p1 (n)  2 p2  p3 . (40)
 3
The speed setting corresponding to the “second” threshold nTr2 can be identified from Equation (40), by satisfying the condition:

M ntr 2   0 .

Starting point for II quadrant

x˙
x     x   2x  kWL   x
  x     x   2x  kWL   x
 
x

Starting point for IV quadrant

Figure 17: Initial Conditions for Calculation of the Boundary

The Second Generation of Intact Stability Criteria: An Overview of Development

Level 1 Vulnerability Criteria for Surf-riding and Broaching-to
The “second” surf-riding threshold is already used in MSC.1/Circ. 1228 as operational guidance for avoiding the danger of broaching.
First introduced in 1995 by MSC.707, this threshold was based on phase-plane analyses of surging and surf-riding for a number of
ships, using a wave steepness of 1/10. The analysis resulted in the following formula:

1.8 L
VS  , kn, (41)
cos 180   

where L is length of the ship,  is wave heading (0° is head waves). Assuming following seas ( = 180°) and transforming Equation
(41) into the form of length-based Froude number yields:

1.8  0.51444
Fn   0.296  0.3 . (42)
g

Taking advantage of the experience with the application of MSC.1/Circ. 1228, it makes sense to consider Equation (42) as a level 1
vulnerability criterion. Thus, ships with service speeds exceeding Fn = 0.3 should be subject to the level 2 vulnerability check. In
general, this is supported by calculations of the second threshold using strip theory for surging force (Equations (34)–(36), with further
direct analytical detection by the Melnikov method (Spyrou, 2006) or by numerical integration along invariant manifolds with
Newton’s method (Umeda, et al., 2007). Umeda & Yamamura (2010) reported that the second threshold could be slightly below
Fn = 0.3 for certain types of vessels, though further study by Umeda, et al. (2011) did not support this. One reason why calculations
can sometimes show the second threshold below Fn = 0.3 is the overestimation of surging forces for some particularly fine ship forms.

Assuming the Froude number as a vulnerability criterion and 0.3 as a standard, it is necessary to recognize that this criterion’s physical
background is related to a wave steepness of 1/10. This wave steepness is quite high and the likelihood of encountering a long and
steep wave is less than the likelihood of encountering a short and steep wave. Accepting Fn = 0.3 as an “across-the-board” criterion
and standard may unnecessarily penalize long ships; thus the length of a ship needs to be included in the criterion. Document SLF-
53/3/8 considers vulnerability for surf-riding and broaching-to only for a ship when its length is less than 200 m.

The critical wave/wave group approach (Themelis & Spyrou, 2007) combines the probabilistic nature of realistic waves and a
deterministic description of ship dynamics. Umeda & Yamamura (2010) examined estimates of the probability of broaching-to using
ships of different lengths. First, combinations of wave height and wave length leading to dangerous broaching-to were determined by
numerical simulations in regular waves. Then, the probability of encountering these waves was evaluated using the Longuet-Higgins
method (Longuet-Higgins, 1983). Finally, the probability of broaching-to was estimated as the probability of encountering a wave
capable of causing broaching-to (Umeda, et al., 2007). The simultaneous increase of the wave length and wave height led to a decrease
of the probability of broaching-to. Similar conclusions were reached by Peters, et al. (2011).

Results from applying the proposed criteria to the sample ships are given in Tables 7 and 8. The service speed is assumed to be 1 knot
below the design speed. Only two ships from Table 7 and four ships from Table 8 are shown to have vulnerability to surf-riding and
broaching-to, as their length is below 200 m and the Froude number, corresponding to service speed, exceeds 0.3. The fishing vessels
and naval combatants from Table 8 are generally known for their vulnerability to surf-riding. The tugs from Table 7 show vulnerability
to surf-riding because they are short and powerful. However, because the operation of these relatively small tugs in open sea is
unlikely, surf-riding and broaching-to is not a hazard for them.

Table 7: Sample Calculations for Surf-riding Level 1 Criteria, based on Ships from SLF 53/INF.10 Annex 10
L, V, Fn > V, Fn >
Ship Description m kn Fn 0.3 ? Ship Description L, m kn Fn 0.3 ?
Bulk Carrier 5500 DWT 190 14 0.15 N Containership >500 TEU 135 18 0.24 N
Bulk Carrier 180 14 0.16 N Containership >500 TEU 125 18 0.25 N
Containership >10000 TEU 360 25 0.21 N Cruise Vessel 240 22 0.22 N
Containership >10000 TEU 360 24 0.20 N LNG Carrier 1000 cbm 110 16 0.23 N
Containership > 6000 TEU 320 25 0.22 N MPV 135 15 0.20 N
Containership > 6000 TEU 320 25 0.22 N MPV 125 15 0.21 N

The Second Generation of Intact Stability Criteria: An Overview of Development

 Table 7(Cont.): Sample Calculations for Surf-riding Level 1 Criteria, based on Ships from SLF 53/INF.10 Annex 10
L, V, Fn > V, Fn >
Ship Description m kn Fn 0.3 ? Ship Description L, m kn Fn 0.3 ?
Containership > 4000 TEU 250 24 0.24 N MPV 120 17 0.24 N
Containership > 4000 TEU 250 25 0.25 N MPV 7500 DWT 105 13 0.19 N
Containership > 1000 TEU 210 17 0.18 N Tanker 30000 DWT 320 14 0.12 N
Containership > 1000 TEU 200 22 0.24 N Tanker 110 13 0.19 N
Containership > 1000 TEU 170 20 0.24 N Tug 30 12 0.33 Y
Containership > 1000 TEU 160 20 0.25 N Tug 25 11 0.33 Y

Table 8: Sample Calculations for Surf-riding Level 1 and 2 Criteria, based on Ships from SLF 53/INF.10 Annex 5
V, Fn > C2L >
Ship Description L, m kn Fn 0.3 ? C2 C2L 0.01 ?
Bulk Carrier 275 16 0.15 N 0.00 0.00 N
Bulk Carrier 2 145 14 0.18 N 0.00 0.00 N
Containership 1 Post-panamax 323 25 0.22 N 0.00 0.00 N
Containership 2 Post-panamax 376 25 0.20 N 0.00 0.00 N
Containership 3 Post-panamax 330 25 0.22 N 0.00 0.00 N
Containership 4 Panamax 283 25 0.23 N 0.00 0.00 N
Containership 5 C11 Class 262 25 0.24 N 0.00 0.00 N
Fishing Vessel 1 Japanese Purse Seiner 35 17 0.44 Y 0.56 0.285 Y
Fishing Vessel 2 22 14 0.46 Y 0.34 0.175 Y
General Cargo 1 Series 60 CB = 0.7 122 18 0.25 N 0.00 3 10 -9
N
General Cargo 2 C4 Class 161 18 0.22 N 0.00 0.00 N
LNG Carrier 268 18 0.17 N 0.00 0.00 N
Naval Combatant 1 ONR Tumblehome 150 30 0.39 Y 0.20 0.22 Y
Naval Combatant 2 ONR Flared 150 30 0.39 Y 0.20 0.22 Y
Passenger Ship 276 25 0.24 N 0.00 0.00 N
RoPax 137 19 0.25 N 0.00 5 10 -5
N
Tanker 320 14 0.12 N 0.00 0.00 N

Level 2 Vulnerability Criteria for Surf-riding and Broaching-to

Similar to pure loss of stability, the phenomenon of surf-riding is often a single-wave event. Despite the fact that the process of
attraction to a surf-riding equilibrium takes some time, the disappearance of the equilibrium is instantaneous. The vulnerability to surf-
riding can be measured by the percentage of waves capable of generating surf-riding. To do so, the irregular seaway is modeled as a
sequence of sinusoidal waves with random lengths and amplitudes and the statistical weight of each wave is calculated using Equation
(3). The results can be regarded as the probability of occurrence of surf-riding when a ship meets a wave with the assumed wave length
and amplitude.

Direct calculation of the second threshold by Melnikov’s method or a numerical method using the phase plane is considered too
complex for the level 1 vulnerability criterion. However, such level of complexity is consistent with requirements for level 2.

For each wave with a specific length and amplitude associated with a wave spectrum, the speed of the ship is compared with the speed
corresponding to the “second” threshold for surf-riding, calculated by Melnikov’s method. This comparison (using Froude number,
Fn) yields a factor C2ij:

The Second Generation of Intact Stability Criteria: An Overview of Development

 1 if Fn  FnTR ( i , a j )
C 2ij  
0 if Fn  FnTR ( i , a j ) .

The weighted average of the factor, C2ij, which is determined over all of the values of λi and aj, is the key element of the criterion:

N N a
C 2( H S , TZ )  W C 2
i 1 j 1
ij ij . (43)

where criterion C2 is a function of the significant wave height, HS, and the mean zero-crossing period, TZ, since the distribution of wave
numbers and amplitudes used for calculation of statistical weights depends on the spectrum defined using these parameters.

The long-term formulation of the criterion can be found by averaging Equation (43) over the values of significant wave heights and
mean zero-crossing periods using entries of the wave scatter diagram as weights:

C2L  C 2(H

H S TZ
S , TZ ) P H S , TZ  .

where P(HS,TZ) is the statistical frequency of observation of a sea state with significant wave height, HS, and mean zero-crossing
period, TZ. The data for averaging between different sea states can be obtained with the standard wave scatter diagram (i.e. IACS
2001):

N H S , TZ 
PH S , TZ   .
NTot

where N(HS,TZ) is the number of observations of a particular sea state and NTot is the total observations available.

Table 8 contains the results of calculations carried out on a sample population of ships. The value of criterion C2 is calculated for a
single wave-state characterized with HS = 2.5 m, TZ = 8.5 s (short-term criterion). Assuming a standard of 0.01, the susceptibility to
surf-riding and broaching-to is predicted for both fishing vessels and naval combatants. This prediction is generally consistent with
operational experience. The differences between short-term and long-term criteria seem to be minimal. The long-term criteria seem to
be more sensitive as it covers more wave-height/wave-length combinations. The results are also consistent with the level 1
vulnerability criterion.

DIRECT STABILITY ASSESSMENT AND FUTURE OUTLOOK

Tools
Once the vulnerability to a certain mode of stability failure has been established, a direct assessment of dynamic stability is expected to
follow, as defined in the framework of the second generation intact stability criteria (Annex 2 SLF 51/4/1). The objective of the direct
stability assessment is two-fold: one is as a tool for detailed design analysis and the other is as a tool for the development of ship-
specific operational guidance. At the core of direct stability assessment lies a method capable of reproducing ship motions in severe
seas, with sufficient fidelity for sound, technical decision-making.

Considering the current state-of-the art of computational ship hydrodynamics for these problems, general direct stability assessment
options appear to be limited to model tests and fast time-domain simulations. The simulations use potential flow wave-body
hydrodynamic methods and are supplemented by empirical formulations for viscous and vortex forces, which are based on model tests.

The main advantage of the current potential flow codes is that body-nonlinear, hydrostatic and Froude-Krylov forces can be computed
efficiently and accurately. As a result, there is no need for the separation of restoring and exciting forces; strictly speaking, that can be
done only under assumptions of linearity. Keeping in mind that both pure loss of stability and parametric roll are largely driven by
hydrostatic and Froude-Krylov forces, the fidelity of their modeling requires a serious consideration.

The evaluation of radiation and diffraction forces by a potential-flow code involves additional assumptions related to the techniques for
solving the boundary-value problem. Therefore, one may expect an increase in the scatter in the results when benchmarking different
codes.

Problems related to maneuvering in waves are more difficult to solve numerically than problems related to pure loss of stability and
parametric roll. The hull forces in maneuvering depend on vortex phenomena and cannot be easily modeled using potential flow

The Second Generation of Intact Stability Criteria: An Overview of Development

assumptions. Inclusion of empirically-based approximations for the effect of lifting phenomenon in potential flow codes may lead to
significant errors. These errors are partially caused by double counting for wave forces that are implicitly included in the empirical
data and inherently calculated by the potential code. Despite these difficulties, there has been substantial progress in the development
of hydrodynamic models for maneuvering in waves (cf. Yen, et al., 2010). Meanwhile, the modeling of ship maneuvering by a system
of ordinary differential equations remains a mature and practical tool (SLF 53/3/8).

Validation remains an important problem for all tools that might be used in a regulatory framework (Reed, 2008). Additional work is
needed to determine how the scatter in results may influence the application of these codes in a regulatory framework. The answer may
be different for different modes of stability failure. However, this issue remains outside the scope of this paper.

Procedures
Having a numerical tool, even if it is validated, is insufficient for practical application in direct stability assessment. There must be a
process for applying the tool that leads to a conclusion regarding the stability of the vessel under consideration.

The framework of the second generation intact stability criteria (Annex 2 SLF 51/4/1) calls for evaluation of the safety level for each
mode of stability failure. If the safety level is expressed in the form of a long-term probability of a particular mode of failure, it creates
a basis for comparison and a defendable outcome of the direct stability assessment. The importance of establishing good safety level
criteria is that the level of safety of a new ship design can be judged against the safety level of an existing ship with a long history of
safe operation.

Evaluation of long-term probability is straightforward (Sevastianov, 1994) if the short-term probability of failure is available. A short-
term probability of failure is evaluated under the assumption that the environment can be described as a stationary stochastic process.
However, calculation of this probability represents a significant challenge because a stability failure is, fortunately, a rare event. Thus,
the estimate of the probability of failure by direct counting is impractical. The calculation of short-term probability of ship stability
failure is defined in Annex 2 of SLF 51/4/1 as the problem of rarity. Addressing the problem of rarity requires application of special
extrapolation procedures. Each mode of failure may require a specific extrapolation procedure that suits the particular physics of the
phenomenon.

CONCLUSIONS
The objective of this paper is to describe the current state of IMO’s development of second generation intact stability criteria.
Following the completion of the 2008 IS code, the SLF Sub-Committee focused on new criteria that would address the dynamic
stability failures not yet covered, including pure loss of stability, parametric roll, and broaching-to.

The concept of the second generation intact stability criteria is based on a multi-tiered evaluation process. Because the direct stability
assessment of ship-stability performance may incur substantial additional design analysis expense, a vulnerability check needs to be
performed first, to exclude cases where the modes of stability failure are not a concern. Vulnerability checks are performed at two
levels to ensure simplicity and prevent unnecessary conservatism. The level 1 vulnerability check is simple, but conservative, while the
level 2 is less conservative, but involves more calculations.

The level 1 vulnerability criteria for pure loss of stability is based on geometry of the hull, reflecting how dramatic changes of
waterplane increase the likelihood for stability failure. The level 2 vulnerability check is performed with the GZ curve changing in
waves, using different parameters as the criteria.

For parametric roll, the level 1 vulnerability criteria is based on the simplest model of parametric resonance—the Mathieu equation.
Two conditions are used: the magnitude of stability changes in waves, and the encounter-frequency condition based on ship-service
speed. The level 2 vulnerability criterion is based on the roll response under conditions of parametric resonance.

The vulnerability criteria for broaching-to focus on surf-riding, as the latter is usually a prerequisite for broaching-to. The level 1
criterion is formulated as a simple condition for service speed in terms of Froude number and vessel length. The level 2 criterion is
based on the threshold speed for surf-riding under any initial conditions.

The next objective in the development of the second generation intact stability criteria is defining the requirements and procedures
necessary for direct assessment. This is a formidable task. Not only must the most advanced technologies available be used, but they
also need to be available world wide

The Second Generation of Intact Stability Criteria: An Overview of Development

At this moment, advanced potential-flow hydrodynamic codes with empirical models for lifting and viscous forces seem to be the most
appropriate tools for predicting pure loss of stability and parametric roll. Broaching-to is best modeled with a system of ordinary
differential equations until more advanced hydrodynamic codes gain enough maturity for this application.

Development of validation procedures for time-domain simulation tools is another difficult but absolutely necessary task. As direct
stability assessment is meant to be done under realistic conditions, the stochastic nature of the environment needs to be fully
considered. This means that stability failures must be regarded as random events and, since they are rare, the problem of rarity needs to
be addressed with a set of appropriate probabilistic extrapolation procedures. Development and verification of these procedures is just
as important as the development and validation of numerical tools.

Also, an additional mode of intact stability failure, excessive accelerations, needs to be addressed with suitable vulnerability criteria
and direct stability assessment methods.

Once the criteria development is complete, two more stages will be needed. First, standards or acceptance boundaries must be
established for the various criteria. This task will involve agreeing on socially acceptable risks of maritime activity. While this
problem has been tackled in other fields of engineering, it has not yet been adequately addressed as far as intact stability is concerned.

Second, issues of implementation of the second generation intact stability regulations will need to be addressed. Implementation will
rely on careful testing of the new criteria and a comprehensive analysis of the impact of new regulations on exiting and future fleets.

DISCLAIMER
The views and opinions expressed in this paper are solely and strictly those of the authors and do not necessarily reflect those of any
organizations with which the authors have been associated, the national delegations in which the authors may participate, or the
International Maritime Organization’s working and correspondence groups on intact stability.

ACKNOWLEDGMENTS
Part of this work was supported by the USCG Office of Design and Engineering Standards, under the supervision of Mr. J. Sirkar.
Some of the methods applied were based on results of research funded by the Office of Naval Research, under the supervision of Dr. L.
P. Purtell. This support is greatly appreciated.

Part of this work was supported by a Grant-in Aid for Scientific Research of the Japan Society for Promotion of Science (No.
21360427). It was partly carried out as a research activity of the Stability Project of the Japan Ship Technology Research Association
in the fiscal year of 2010 and funded by the Nippon Foundation. The authors express their sincere gratitude to the above organizations.

Further, the authors thank Mr. Ole Hympendahl who performed certain validation work that provided an independent check of
procedures and criteria proposed.

The authors are grateful to both Dr. Art Reed and Ms. Suzanne Reed for their detailed editing that has greatly improved clarity and
readability of the text.

REFERENCES
ABS (2004) Guide for the Assessment of Parametric Roll Resonance in the Design of Container Carriers, American Bureau of
Shipping, Houston, TX, 70 p.
BASSLER, C. C., V. Belenky, G. Bulian, A. Francescutto, K. Spyrou, N. Umeda (2009) A Review of Available Methods for
Application to Second Level Vulnerability Criteria. Proc. 10th Int’l Conf. Stability of Ships & Ocean Vehicles (STAB ’09), St.
Petersburg, Russia, pp. 111–128.
BASSLER, C. C., R. Miller, A. M. Reed & A. Brown (2011) Considerations for Bilge Keel Force Models in Potential Flow
Simulations of Ship Maneuvering in Waves. Proc. 12th Int’l Ship Stability Workshop, Washington, DC, pp. 291–307.
BECK, R. F. & A. M. Reed (2001). Modern computational methods for ships in seaway. Trans. SNAME, 109:1–48.
BELENKY, V. L., ed. (2011) Proc. 12th Int’l Ship Stability Workshop, Washington, DC, viii+420 p.
BELENKY, V. & C. Bassler (2010) Procedures for Early-Stage Naval Ship Design Evaluation of Dynamic Stability: Influence of the
Wave Crest. Naval Engineers J., 122(2):93–106.
BELENKY, V., J. O. de Kat & N. Umeda (2008) Towards Performance-Based Criteria for Intact Stability. Marine Tech., 45(2):101–
123.
BELENKY, V. L. & N. B. Sevastianov (2007) Stability and Safety of Ships: Risk of Capsizing (2nd ed.). SNAME, Jersey City, NJ,
xx+435 p.

The Second Generation of Intact Stability Criteria: An Overview of Development

BELKNAP, W. F. & A. M. Reed (2010) TEMPEST — A New Computationally Efficient Dynamic Stability Prediction Tool. Proc. of
11th Int’l Ship Stability Workshop, Wageningen, the Netherlands, pp. 185–197.
BISHOP, R. E. D. & Price, W. G. (1978) On the truncation of spectra. Int’l Shipbuilding Prog., 25:3–6.
BRUNSWIG, J. & R. Pereira (2006) Validation of Parametric Roll Motion Predictions for a Modern Containership Design. Proc. 9th
Int’l Conf. Stability Ships & Ocean Vehicles (‘06), Vol. 1, Rio de Janeiro, Brazil, pp. 157–168.
BSU (2009) Fatal accident on board the CMV Chicago Express during Typhoon “Hagupit” on 24 September 2008 off the coast of
Hong Kong, German Federal Bureau of Maritime Casualty Investigation, Hamburg, Germany, 64 p. (in German) (available on
www.bsu-bund.de)
BULIAN, G. (2004) Approximate analytical response curve for a parametrically excited highly nonlinear 1-DOF system with an
application to ship roll motion prediction. Nonlinear Analysis: Real World Applications, 5:725–748.
BULIAN, G. & A. Francescutto (2008) SAFEDOR Benchmark on Parametric roll—Brief description of the Simulation Methodology
Employed in the Code SHIXDOF under Development at DINMA, Internal Technical Report, Department DINMA, U. Trieste,
Trieste, Italy.
BULIAN, G. & A. Francescutto (2010) A simplified regulatory-oriented method for relative assessment of susceptibility to parametric
roll inception at the early design stage. Proc. 4th Int’l Maritime Conf. on Design for Safety and 3rd Workshop on Risk-Based
Approaches in the Marine Industries - Part I, Trieste, Italy, pp. 93–106.
BULIAN, G. & A. Francescutto (2011) Considerations on Parametric Roll and Dead Ship Condition for the Development of Second
Generation Intact Stability Criteria. Proc. 12th Int’l Ship Stability Workshop, Washington, DC, pp. 7–18.
DE KAT, J. O., R. Brouwer, K. McTaggart &, W. L. Thomas (1994). Intact Ship Survivability in Extreme Waves: New Criteria from
Research and Navy Perspective. Proc. 5th Int’l Conf. Stability of Ships & Ocean Vehicles (STAB ’94), Vol. 1, Melbourne, FL, 26
p.
DEGTYAREV, A., ed. (2009) Proc. 10th Int’l Conf. Stability of Ships & Ocean Vehicles (STAB ’09). St. Petersburg, Russia, 754 p.
FRANCE, W. M, M. Levadou, T. W. Treakle, J. R. Paulling, K. Michel & C. Moore (2003). An Investigation of Head-Sea Parametric
Rolling and its Influence on Container Lashing Systems. Marine Tech.¸ 40(1):1–19.
FRANCESCUTTO, A. (2004) Intact Ship Stability—The Way Ahead. Marine Tech., 41:31–37.
FRANCESCUTTO, A. (2007) Intact Stability of Ships-Recent Developments and Trends. Proc.PRADS ’07, Houston, TX.
GRAFF, W. & E. Heckscher (1941) Widerstands und Stabilitäts-Versuche mit drei Fischdampfermodellen. Werft-Reederei-Hafen,
22:115–120.
GRIM, O.(1951) Das Schiff von Achtern Auflaufender. Jahrbuch der Schiffbautechnischen Gesellschaft, 4:264–287.
HASHIMOTO, H. (2009) Pure Loss of Stability of a Tumblehome Hull in Following Seas. Proc. 19th Int’l Offshore and Polar Engin.
Conf., Vol. 3, Osaka, Japan, pp. 717-721.
HASHIMOTO, H., N. Umeda & A. Matsuda (2011) Broaching prediction of a wave-piercing tumblehome vessel with twin screws and
twin rudders. J. Marine Sci. & Tech. 14 p. (DOI 10.1007/s00773-011-0134-1)
HAYASHI, C. (1985) Nonlinear Oscillations in Physical Systems, Princeton University Press, Princeton, NJ, xii+392 p.
IACS (2001) Standard Wave Data. International Association of Classification Societies, Ltd., Recommendation No. 34, Rev. 1,
London, UK, 4 p.
IKEDA, Y., M. Kashiwagi & N. Umeda, eds. (2008) Proc. 6th Osaka Colloquium on Seakeeping and Stability of Ships, Osaka, Japan.
IMO (2009) International Code on Intact Stability, 2008, IMO, London, UK, v+160 p.
IMO MSC.1/Circ.1200 (2006) Interim Guidelines for Alternative Assessment of the Weather Criterion, London, UK, 17 p.
IMO MSC.1/Circ.1227 (2007) Explanatory Notes to the Interim Guidelines for Alternative Assessment of the Weather Criterion,
London, UK, 23 p.
IMO MSC.1/Circ.1228 (2007) Revised Guidance for Avoiding Dangerous Situations in Adverse Weather and Sea Conditions, London,
UK, 8 p.
IMO MSC.1/Circ.1281 (2008) Explanatory Notes to the International Code on Intact Stability, 2008, London, 30 p.
IMO MSC.707 (1995) Guidance to the Master for Avoiding Dangerous Situations in Following and Quartering Seas. London, UK. 14
p.
IMO Res. A.167(ES.IV) (1968) Recommendation On Intact Stability for Passenger and Cargo Ships Under 100 metres in Length,
London, UK, 15 p.
IMO Res. A.562(14) (1985) Recommendation on a severe wind and rolling criterion (weather criterion) for the intact stability of
passenger and cargo ships of 24 metres in length and over, London, UK, 5 p.
IMO Res. A.749(18) (1993) Code on Intact Stability for all Types of Ships Covered by IMO Instruments, London, UK, 77 p.
IMO Res. MSC.269(85) (2008) Adoption of Amendments to the International Convention for the Safety of Life At Sea, 1974, as
amended, London, UK, 11 p.
IMO Res. MSC.270(85) (2008) Adoption of Amendments to the Protocol of 1988 Relating to the International Convention on Load
Lines, 1966, as amended, London, UK, 2 p.
IMO SLF 48/4/12 (2005) On the Development of Performance-Based Criteria for Ship Stability in Longitudinal Waves, Submitted by
Italy, IMO, London, UK, 6 p.
IMO SLF 48/21 (2005) Report to Maritime Safety Committee, London, UK, 65 p.

The Second Generation of Intact Stability Criteria: An Overview of Development

IMO SLF 50/4/4 (2007) Framework for the Development of New Generation Criteria for Intact Stability, submitted by Japan, the
Netherlands and the United States, London, UK, 6 p.
IMO SLF 51/4/1 (2008) Report of the Intercessional Correspondence Group on Intact Stability, Submitted by Germany, London, UK,
25 p.
IMO SLF 51/INF.4 (2008) Development of Performance-Based Intact Stability Criteria, Submitted by the United States, London, UK,
2 p.
IMO SLF 51/INF.6 (2008) Research outcomes for new generation intact stability criteria. Submitted by the Royal Institution of Naval
Architects (RINA), London, UK, 4 p.
IMO SLF 52/3/1 (2009) Report of the Intercessional Correspondence Group on Intact Stability, Submitted by Japan, London, UK, 19 p.
IMO SLF 52/INF.2 (2009) Information Collected by the Intercessional Correspondence Group on Intact Stability, Submitted by Japan,
London, UK, 138 p.
IMO SLF 53/3/1 (2010) Report of the Intercessional Correspondence Group on Intact Stability, Submitted by Japan, London, UK, 7 p.
IMO SLF 53/3/3 (2010) Activities of the Dynamic Stability Task Group of the Society of Naval Architects and marine Engineers,
Submitted by the Royal Institution of Naval Architects (RINA), London, UK, 2 p.
IMO SLF 53/3/7 (2010) Comments on Level 1 Vulnerability Criteria for Parametric Rolling, Submitted by the United States, London,
UK, 5 p.
IMO SLF 53/3/8 (2010) Comments on Proposed Criteria for Surf-riding and Broaching, Submitted by Japan and the United States,
London, UK, 6 p.
IMO SLF 53/3/9 (2010) Comments on documents SLF 53/3/1 and SLF 53/INF.10. Submitted by Italy, London, 4p.
IMO SLF 53/19 (2011) Report to Maritime Safety Committee, London, UK, 43 p.
IMO SLF 53/INF.8 (2010) Sample Calculations on the Level 2 Vulnerability Criteria for Parametric Roll, Submitted by Sweden,
London, UK, 18 p.
IMO SLF 53/INF.10 (2010) Information Collected by the Correspondence Group on Intact Stability, Submitted by Japan, London, UK,
151 p.
IMO SLF 53/WP.4 (2011) Report of the Working Group (Part 1), London, UK, 20 p.
JENSEN, J. J. (2007) Efficient estimation of extreme non-linear roll motions using the first-order reliability method (FORM). J. Marine
Sci. & Tech, 12:191-202.
KAN, M. (1990) A Guideline to Avoid the Dangerous Surf-riding. Proc. 4th Int’l Conf. Stability of Ships & Ocean Vehicles,
University Federico II of Naples, Naples, Italy, pp. 90–97.
KEMPF, G. (1938) Die Stabilitätsbeanspru-chung der Schiffe durch Wellen und Schwingungen. Werft-Reederei-Hafen, 19:200–202.
KOBYLINSKI, L. K., ed. (2009) Proc. Int’l Workshop on Dynamic Stability Consideration in Ship Design, Foundation for Safety of
Navigation and Environment Protection, Ilawa, Poland. (www.ilawashiphandling.com.pl/ workshop2009/workshop2009.htm)
KOBYLINSKI, L. K. & S. Kastner (2003) Stability and Safety of Ships: Regulation and Operation. Elsevier, Amsterdam, 454 p.
KROEGER, H-P (1986) Rollsimulation von schiffen im seegang. Schiffstechnik, 33:187–216.
LIN. W. M. & D. K. P. Yue (1990) Numerical Solutions for Large Amplitude Ship Motions in the Time-Domain. Proc. 18th Symp. of
Naval Hydro., Ann Arbor, MI, pp. 41–66.
LONGUET-HIGGINS, M. S. (1957) The Statistical Analysis of a Random, Moving Surface. Phil. Trans. Royal Soc. London, A,
249(962):321–387.
LONGUET-HIGGINS, M. S. (1976) On the Nonlinear Transfer of Energy in the Peak of a Gravity-Wave Spectrum: A Simplified
Model. Proc. Royal Soc. London A, 347:311–328.
LONGUET-HIGGINS, M. S. (1983) On the Joint Distribution of Wave Periods and Amplitudes in a Random Wave Field. Proc. Royal
Soc. London A, 389:241–258.
LONGUET-HIGGINS, M. S. (1984) Statistical Properties of Wave Groups in a Random Sea State. Phil. Trans. Royal Soc. London, A,
312(1521):219–250.
MAKOV, Y. (1969) Some Results of Theoretical Analysis of Surf-riding in Following Seas. Trans. Krylov Soc., Leningrad, 126:124-
128. (in Russian)
MARITIME New Zealand (2007) Incident Report Heavy Weather/Cargo Shift Aratere 3 March 2006. Maritime New Zealand
Investigation Report 06-201, vi+57 p.
MATUSIAK, J. (2000) Two-stage Approach to Determination of Non-linear Motions of ship in Waves. Proc. 4th Osaka Colloquium
on Seakeeping Performance of Ships, Osaka, Japan.
NEVES, M. A. S., J. E. M. Vivanco & C. A. Rodríguez (2009) Nonlinear Dynamics on Parametric Rolling of Ships in Head Seas. Proc.
10th Int’l Conf. Stability of Ships & Ocean Vehicles, St. Petersburg, Russia, pp. 509–520.
OH, I. G., A. H. Nayfeh & D. T. Mook (2000) A theoretical and experimental investigation of indirectly excited roll motion in ships.
Phil. Trans. Royal Soc. London. 358(1771):1853–1881.
PAULLING, J. R. (1961) The Transverse Stability of a Ship in a Longitudinal Seaway. J. Ship Res., 4(4):37–49.
PAULLING, J. R., S. Kastner & S. Schaffran (1972) Experimental studies of capsizing of intact ships in heavy seas. US Coast Guard,
Technical Report, 58 p. (Also IMO Doc. STAB/7, 1973)

The Second Generation of Intact Stability Criteria: An Overview of Development

PAULLING, J. R., O. H. Oakley & P. D. Wood (1974) Ship motions and capsizing in astern seas. Proc. 10th Symp. Naval Hydro.,
Boston, MA, pp. 532–549.
PAULLING, J. R., O. H. Oakley & P. D. Wood (1975) Ship capsizing in heavy seas: the correlation of theory and experiments. Paper
4.3, Proc.1st Int’l Conf. Stability of Ships & Ocean Vehicles (STAB ’75), Glasgow, 19 p.
PAULLING J. R. & R. M. Rosenberg (1959) On Unstable Ship Motions Resulting from Nonlinear Coupling. J. Ship Res., 3(1):36–46.
PEREIRA, R. (1988) Simulation nichtlinearer Seegangslasten. Schiffstechnik, 35:173-193.
PETERS, W. S., V. Belenky & C. Bassler (2010) On Vulnerability Criteria for Righting Lever Variations in Waves. Proc. 11th Int’l
Ship Stability Workshop, Wageningen, the Netherlands. pp. 7–16.
PETERS, W. S., V. Belenky, C. Bassler & K. Spyrou (2011) On Vulnerability Criteria for Parametric Roll and Surf-riding. Proc. 12th
Int’l Ship Stability Workshop, Washington, DC, pp. 1–6.
RAHOLA, J. (1939) The Judging of the Stability of Ships and the Determination of the Minimum Amount of Stability Especially
Considering the Vessel Navigating Finnish Waters. PhD Thesis, Technical University of Finland, Helsinki, viii+232 p.
REED, A. M. (2008) Discussion of: Belenky, V., J. O. de Kat & N. Umeda (2008) Towards Performance-Based Criteria for Intact
Stability. Marine Tech., 45(2):122–123.
RICE, S. O. (1944/45) Mathematical Analysis of Random Noise. Bell System Techn J., 23(3):282–332, 24(1):46–156.
SADAT-HOSSEINI, H., P. Carrica, F. Stern, N. Umeda, H. Hashimoto, S. Yamamura & A. Mastuda (2011) CFD, system-based and
EFD study of ship dynamic instability events: Surf-riding, periodic motion, and broaching. Ocean Engin., 38:88–110.
SANCHEZ, N. E. & A. H. Nayfeh (1990). Nonlinear rolling motions of ships in longitudinal waves. Int’l Shipbuilding Prog.,
37(411):247–272.
SCHREUDER, M. (2005) Time Simulation of the Behaviour of Damaged Ships in Waves. Department of Shipping and Marine
Technology, Chalmers University of Technology, Göteborg, Sweden, 49 p.
SEVASTIANOV, N. B. (1994) An Algorithm of Probabilistic Stability Assessment and Standards. Proc. 5th Int’l Conf. Stability of
Ships & Ocean Vehicles (STAB ‘94), Vol. 5, Melbourne, FL, 12 p.
SHIGUNOV, V. (2009) Operational Guidance for Prevention of Container Loss. Proc. 10th Int’l Conf. Stability of Ships & Ocean
Vehicles (STAB ‘09), St. Petersburg, Russia, pp. 473–482.
SHIGUNOV, V. & R. Pereira (2009) Direct Assessment Procedure and Operational Guidance for Avoidance of Cargo Loss and
Damage on Container Ships in Heavy Weather, International Workshop on Dynamic Stability Considerations in Ship Design
(DSCSD), 11 p. (www.ilawashiphandling. com.pl/workshop2009/workshop2009.htm)
SHIGUNOV, V., H. Rathje, O. El Moctar & B. Altmayer (2011) On the Consideration of Lateral Accelerations in Ship Design Rules.
Proc. 12th Int’l Ship Stability Workshop, Washington, DC, pp. 27–36.
SÖDING, H. (1982) Leckstabilität im Seegang. Report No. 429, Institut für Schiffbau, Hamburg, Germany, 69 p.
SPANOS, D. & A. Papanikolaou (2006) Numerical Simulation of Parametric Roll in Head Seas. Proc. 9th Int’l Conf. Stability of Ships
& Ocean Vehicles (STAB ‘06), Vol. 1, Rio de Janeiro, Brazil, pp. 169–180.
SPYROU, K. J. (1996) Dynamic Instability in Quartering Seas: The Behavior of a Ship During Broaching. J. Ship Res., 40(1):46–59.
SPYROU, K. J. (1997) Dynamic Instability in Quartering Seas-Part III: Nonlinear Effects on Periodic Motions. J. Ship Res.,41(3):210–
223.
SPYROU, K. J. (2005) Design Criteria for Parametric Rolling. Oceanic Engin. Int’l, 9(1):11–27.
SPYROU, K. J. (2006) Asymmetric Surging of Ships in Following Seas and its Repercussion for Safety. Nonlinear Dynamics, 43:149–
172.
ST. DENIS, M. (1980) Some Comments on Certain Idealized Variance Spectra of the Seaway Currently in Fashion. J. Ship Res.,
24(4):271–278.
THEMELIS, N. & K. J. Spyrou (2007) Probabilistic Assessment of Ship Stability. Trans. SNAME, 117:181–206.
UMEDA, N., S. Izawa, H. Sano, H. Kubo & K. Yamane (2011) Validation & Verification on Draft New Generation Intact Stability
Criteria. Proc. 12th Int’l Ship Stability Workshop, Washington, DC, pp. 19–26.
UMEDA, N., M. Shuto & A. Maki (2007) Theoretical Prediction of Broaching Probability for a Ship in Irregular Astern Seas. Proc. 9th
Intl. Ship Stability Workshop, Hamburg, Germany, 7 p.
UMEDA, N. & S. Yamamura (2010) Designing New Generation Intact Stability Criteria on Broaching Associated with Surf-Riding.
Proc. 11th Int’l Ship Stability Workshop, Wageningen, the Netherlands, pp. 17–25.
WALREE, F. van, ed. (2010) Proc. 11th Int’l Ship Stability Workshop, Wageningen, the Netherlands.
WATANABE, Y. (1934) On the Dynamic Properties of the Transverse Instability of a Ship due to Pitching. J. Society of Naval Arch.
Japan, 53:51–70. (in Japanese)
YEN, T.-G., S. Zhang, K. M. Weems & W.-M. Lin (2010) Development and Validation of Numerical Simulations for Ship
Maneuvering in Calm Water and in Waves. Proc. 28th Symp. Naval Hydro., Pasadena, CA, 28 p.

The Second Generation of Intact Stability Criteria: An Overview of Development

APPENDIX: TRANSITION-BASED BOUNDARY FOR PARAMETRIC ROLL
In principle, the effect of transients on the initiation of parametric roll can be incorporated into the frequency condition. Using the
theoretical background in Hayashi (1985), following Spyrou (2005), transient effects for conditions that are close to e, but not
necessarily equal to 2m, were used in SLF 53/3/9 and in Bulian & Francescutto (2011).

As a demonstration, the solution of the roll equation represented by Equation (13) is considered for a specific pair of initial conditions
that makes the arbitrary constant, C1, equal to zero (this leads to a theoretical "worst case" solution) with a very simple exponential
growth of the envelope:

e
 t  
(t )  C 2 e  t e 2 sin e t    .
 2 
Starting from this form of the solution, it is possible to determine the magnitude of parametric excitation corresponding to a specified
amplification factor, f, in n oscillations. Assuming that /m<<1, such parametric excitation can be approximated as:

2 2 2
 1 e2     T   e 
h  2 1       . (44)
 4 2      
 m   m   m 

Here, T is an additional “allowance” for the transient effect defined as:

ln f  | e |
T  . (45)
4n
By comparing Equation (44) with Equation (25), one can see that the difference between them is the transient effect. This effect is
shown in Figure 18; the transition boundary shows the combinations of parametric excitation and frequency ratio, where the parametric
roll grows faster than the specified values given the specific initial conditions. The initial conditions are the reason why the “transition
curve” in Figure 18 lies below the boundary of the criterion (21) since it represents a “theoretical worst case scenario.”

In the particular case of e  20 , according to Equations (44) and (45) we obtain:

ln f  4
h2  .
n m

Indeed, while the condition of Equation (21) is based on a particular pair of initial conditions (deemed to be realistic), the curve
specified by Equation (44), and as represented in Figure, represents a “theoretical worst case” scenario.

0.8 h
Transition boundary
f=5, n=4, =0.03 m
0.6

0.4 Standard for

criterion
(37)
0.2 Steady state
Steady state =0.0
=0.03 m e/m
1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4

Figure 18: Effect of Transient Behavior on the Boundary for Parametric Roll

The Second Generation of Intact Stability Criteria: An Overview of Development

 Discussion
Dr. Philippe Corrignan, Visitor
This paper presents a very useful and comprehensive review of the developments of the second generation intact stability criteria after
SLF53, with some complementary technical details with respect to the papers submitted to SLF. As a matter of fact, one or several
vulnerability criteria definitions have been proposed today for each level of each phenomenon and all are based on a large number of
assumptions. Thus, it is important, in order to help their understanding by all the IMO actors who will have to comment, test and,
ultimately, adopt them, to present these criteria, the way they can be calculated, and to clearly state the assumptions behind them, in
one as comprehensive and standalone document as possible. The present paper is a good basis in this respect.

Some proposals of level 1 and 2 vulnerability criteria use a probabilistic approach, considering wave length and amplitude distribution
for a given sea state, and, for some, a long term calculation using wave scatter diagrams. Such an approach is very interesting insofar as
the standards can be defined in terms of probability of occurrence, which can then be defined consistently for the vulnerability criteria
and for the direct assessment approach. The use of short and long term calculations is, however, not systematic. It would be interesting
to know the opinion of the authors on the possibility of generalizing such a probabilistic approach to all phenomena and both
vulnerability levels.

Whereas dead ship condition, pure loss of stability, and parametric rolling seem to be well-described in current proposals, surf-riding
and broaching-to seem to be more difficult to comprehend and translate into suitable criteria. Firstly, it is the only phenomenon for
which the level 2 criterion is not fully described. The short and long term formulations are given, but the core calculation of the second
threshold, using Melnikov’s method or other numerical method using the phase plane, is not described. These methods are not
straightforward and not so wide spread in the hydrodynamics community, so it would be very beneficial to describe the calculation
methodology, as for the other phenomena. Secondly, from an approval point of view, surf-riding level 2 vulnerability criteria require
the knowledge of the calm water resistance and thrust curves, which are not standard data available to regulatory bodies today, and are
generally considered as confidential by ship designers and shipyards. It would be interesting to know the authors’ opinion on this
aspect.

Finally, the multi-tiered approach allows for direct stability assessment in case levels 1 and 2 criteria fail. The crucial need for
validation of such a direct assessment tool is very well highlighted in the present paper. However, I would like to raise another specific
point. It is my understanding that if, at this stage, the risk of stability failure to the considered mode cannot be solved by design
modifications, then the ship can be approved provided suitable operational guidance are developed by the direct assessment tool. It is
expected that such guidance will require the knowledge of encountered environmental conditions, in particular wave height and length.
However, encountered wave conditions are usually not ideal/theoretical ones, and it can be very difficult for people onboard to estimate
accurately the wave parameters required by the guidance, whereas, again, the ship has been proved to be particularly sensitive to one or
several stability failure modes. Does this mean that dedicated wave measurement equipment will have to be developed, approved, and
installed onboard such vulnerable ships? Again, the opinion of the authors on this specific point would be welcome.

Dr. Jan de Kat, Fellow

The authors are to be congratulated on providing a timely update on the development of intact stability criteria. There clearly lies a
challenge in arriving at criteria that cover the essential physical elements while at the same time being applicable in a practical sense
and able to distinguish safe from unsafe ships. My comments aim to highlight some practical aspects.

Concerning the background of the various vulnerability criteria, the paper could provide some more detail on assumptions related to
underlying physics and wave mechanics. For the Level 1 Pure Loss of Stability it is not clear what the associated wave characteristics
are; presumably the wave length is taken equal to ship length, what is the corresponding steepness, and does it depend on the length?
For Level 2 it is mentioned that a series of regular waves and steepness values can be used, but is there any guidance on selecting the
relevant sea states? If the wavelength is taken equal to the ship length as the worst case scenario, this would be rather unrealistic for
vessels over e.g. 300 m length. Similarly, could the authors comment on how the wave characteristics should be selected in the case of
Level 1 and 2 Parametric Roll?

With regard to broaching-to, the simple Froude number criterion captured by Equation (42) was based on model test observations,
some of which were reported by De Kat and Thomas (1998). Both Level 1 and 2 surfriding and broaching criteria provide a certain
level of vulnerability to surf-riding based on selected ship speed versus wave celerity. It is not clear how to distinguish acceptable from

The Second Generation of Intact Stability Criteria: An Overview of Development

unacceptable levels, especially in random seaways. Neither will these criteria provide guidance as to how to design a vessel to avoid
broaching or improve course controllability in a seaway, as the only parameter involved is ship speed.

An important part of the criteria development process is their validation and applicability to a variety of ship types. Herewith I would
like to highlight some issues from the perspective of containership stability. From the results shown in Tables 1 through 6 one could
infer that all containerships considered are likely to experience loss of stability and parametric roll and a subsequent Level 3 analysis is
warranted. This picture seems overly conservative and may be based on assumptions that do not fully consider actual operational
conditions. Containerships operate at a variety of loading conditions and speeds, and typically at lower speeds and at lower drafts than
the design point. An example of distribution of draft and speed is shown below for a modern post-Panamax vessel designed for a
maximum speed of 25 knots. In addition, the KG can vary considerably per draft during an operational life cycle, resulting in a large
range of GM values per operating condition.

This raises the question as to how to select relevant ship conditions for validating and applying vulnerability criteria. Clearly the
example vessel very rarely operates at its maximum design speed and mostly operates at drafts less than the maximum draft of 13 m. In
addition, modern post-Panamax vessels tend to operate at relatively high GM levels, where the GM could be well in excess of and
never below 3 or 4 m. Certainly there are vessel classes where the operating GM can be below 1 m, but for example selecting a
10,000+ TEU vessel with GM = 0.8 m comes across as unrealistic. As part of the Level 2 parametric roll assessment, an allowable roll
angle of 20 degrees is suggested by the authors. It should be noted that roll angles between 20 and 30 degrees are not uncommon for
certain containerships (for instance related to roll excitation in stern quartering seas in lightly loaded condition) and that these vessels
can operate without any problems under these conditions. The main engine can continue functioning up to large roll amplitudes, albeit
not at 100% MCR; a 2-stroke engine can operate at lower load levels with increasing roll amplitudes.

ADDITIONAL REFERENCES
J.O. de KAT & W.L. THOMAS Extreme rolling, broaching and capsizing – model tests and simulations of a steered ship in waves,”
Proc. 22nd Symposium on Naval Hydrodynamics, Washington, D.C., August 1998.

Figure D-1: Operational profile of post-Panamax container vessel

Prof. Alberto Francescutto, Member

The authors are to be congratulated for the very interesting and timely paper which presents a kind of international state-of-the-art of
the situation concerning the development of Second Generation Intact Stability Criteria at IMO.

I would start from an important missing reference in the very brief historical introduction. It is internationally acknowledged that
Rahola’s doctoral thesis was the starting point for the General Criterion of Stability (IMO Res. A.167); at the same time, it is also
internationally acknowledged that Pierrottet was the “father” of the Weather Criterion (and not only), with his paper presented at the

The Second Generation of Intact Stability Criteria: An Overview of Development

Institution of Naval Architects in 1935 (Pierrottet, 1935). I would in particular call the attention of the authors on the discussion
contained in the paper and, in particular to the answer of Prof. Pierrottet to the Chairman Sir Archibald Denny:

“To Sir Archibald Denny I would say that I think the problem of stability is rather neglected by ship designers. I can see danger in his
recommendation of empirical, rather than scientific, methods. If the proportions of bridges across rivers were decide empirically, I am
sure that sooner or later there would be many disasters. The limits of the field over which empirical methods can safely be applied are
very vague. It is my opinion, therefore, that no effort should be spared to study scientifically the stability of ships and to ensure that
designers do not neglect its consideration. I am rather doubtful, moreover, if this object can be attained without the application of
binding regulations.…”

It is important to recall a subject of current interest in the frame of the development of the Second Generation Intact Stability Criteria:
the question of the balance between “simplicity” and “discriminating power” involved in the adopted multi-tier approach. In view of
the present availability of conceptual and software resources, there seems to be often an exaggerated attention to the requirement of
simplicity in the first tier, with the drawback of making too often necessary the recourse to the second level check.

The comments of the authors on this point would be welcome.

ADDITIONAL REFERENCES
PIERROTTET, E., (1935) “Standards of Stability for Ships”, Trans. Institution of Naval Architects, Vol. 77, pp. 208-222.

Mr. Bruce Hutchison, Life Fellow

I congratulate the authors on a paper of broad scope, reporting on a topic of great importance. Intact stability, structural adequacy, and
damaged stability are arguably the topics of greatest importance to safety, of which intact stability has primacy when it comes to
preserving life. The authors categorize intact dynamic stability into three broad topics: pure loss of stability in waves, parametric (aka
subharmonic) resonance, and broaching-to. Excessive acceleration is also acknowledged as a concern, but not further elaborated on in
the present paper. A particularly admirable feature of this paper is the structured, hierarchical approach to determining, for any
particular vessel, its vulnerability to dangerous behavior belonging to any of the three considered classes of dynamic stability ‘failure’.
Especially valuable is the simplicity of the lower levels of vulnerability checks, as these make otherwise daunting and complex
problems accessible to a broader class of professionals, and these low level checks will suffice for the majority of vessels.

Do the authors regard the three categories of intact stability (augmented by excessive acceleration) as sufficient or complete? I
particularly wonder about certain classes of smaller vessels with water on deck, for instance fishing vessels; see for example Caglayan
and Storch (1982), Adee et al. (1986), Pantazopoulos (1988), Shin et al. (2003), Belenky et al. (2002), Falzarano and Laranjinha
(2002), and Lee et al. (2006). I am sure that the authors are aware of even more relevant technical literature.

This paper presents surf riding as a proxy for broaching-to. Is surf riding a “necessary” condition? Are there cases of broaching-to that
are not associated with surf riding?

It is noted that the authors’ proposed Level 1 and Level 2 vulnerability checks are for surf riding, not broaching-to, and that they do not
consider any aspect of the maneuvering capability of the vessel being evaluated. Is surf riding a “sufficient” condition to indicate
vulnerability to broaching-to? Or are there cases of surf riding vessels that are completely controllable in yaw and, hence, not
vulnerable to broaching-to?

Tugboats are identified in the paper as showing vulnerability to surf-riding and then dismissed under the argument that “operation of
these relatively small tugs in open sea is unlikely”. I would offer a slightly different perspective and explanation. 30 m tugs engage
in exposed coastwise tows, and tugs only slightly larger (e.g., 38 m length and above) engage in long-distance ocean towing on voyages
of global scale. The likely explanation is that the towline acting through the towline tie-down significantly stabilizes the tug in yaw
such that broaching does not occur. This explanation applies to vessels engaged in towing activities. Tugs and similar sized vessels
(e.g., research, fishing, or supply vessels) when not engaged in towing may well be vulnerable to surf-riding, and by extension
broaching.

The authors introduce equivalent linear damping, E, in their Equation 33 in the context of parametric roll. Equivalent linear damping
was also invoked by MacMaster and Thompson (1994) in their development of a simple design formula, Ak/V ≈ 2E, relating the ratio
of wave slope, Ak, and the angle of vanishing stability, V. E, is understood to be the equivalent linear damping to the full amplitude of
the nonlinear (large angle) roll. I think that it would be safe to say that we know rather little about damping out to very large angles.
Deck edges submerge, bilges emerge and, as one approaches V, the roll cycle dilates. Much of what we do know about roll damping

The Second Generation of Intact Stability Criteria: An Overview of Development

comes from decay tests in calm water. When one considers that this may be occurring in short-crested seas, the question becomes even
more vexing. I believe that we need considerable fundamental research on this point, and I harbor the hope that we will ultimately
validate RANS CFD methods as a way to predict roll damping to large angles in waves.

The present paper does not delve deeply into the direct assessment methods that the authors envision for vessels found to be vulnerable
through Levels 1 and 2 screening evaluations, but the authors do address their vision and the challenges they foresee, and those
challenges are formidable.

I am confident that the authors are aware of the reductions in survivability associated with broken symmetry. Presumably, any direct
assessment procedure would impose requirements for broken symmetry due either to asymmetric loading (and initial list) or the heel
bias resulting from wind.

The present paper makes mention of Melnikov’s method as applied to surf riding. Many authors have addressed the fractal erosion of
the safe basin for ship stability, and Melnikov’s method has been advanced as a lower bound predictor of the onset of chaotic behavior.
As a lower bound predictor, presumably, Melnikov’s method is conservative. Anticipating that Melnikov’s method may be advanced
as part of any new second-generation stability regime, several questions present themselves. Is the level of conservatism associated
with Melnikov’s method reasonable, or is it excessive? Is the level of conservatism consistent and reliable, or is it non-uniform such
that some vessels might be unequally penalized with excessive conservatism?

I am particularly keen to learn more of the authors’ vision for addressing the problem of rarity. The authors state that “the problem of
rarity needs to be addressed with a set of appropriate probabilistic procedures.” What might those procedures be? One might imagine
fast time domain simulations applied to carefully selected sets of ‘design’ wave sequences and, perhaps also, application of random
initial conditions.

It seems very much the current fashion to develop linear wave profiles (x, t) as ‘design’ wave sequences. It must be observed that for
the dynamic stability problem, however, the wave profile is a proxy for the nonlinear roll response,.NL If one prefers, it is equally easy
to find ‘design’ wave sequences that maximize the linear roll response, L, but this only substitutes the linear roll process as a proxy in
lieu of the linear wave. In ship structures, the use of the linear wave profile or the linear wave bending moment BML as proxies for
primary structural loading, such as the nonlinear wave bending moment, BML, has shown considerable promise and success (see, for
example, the paper by Kim, Engle, and Troesch on this same Annual Meeting program). But this success derives in no small part from
the weakly nonlinear character of BML. By way of contrast, the problem of ship dynamic stability is strongly nonlinear, manifesting
behaviors such as subharmonic resonance, broaching-to, and capsizing.

Even when using all of our arts to devise the most revealing program of inquiry, the fractal character of the safe basin together with
these other strong nonlinear response possibilities diminishes one’s sense that any limited program of fast time-domain simulations can
ascertain with known confidence the ‘safe’ or ‘non-safe’ nature of any particular ship.

In closing, I would like to emphasize some of the authors’ own observations regarding the challenges ahead. “Not only must the most
advanced technologies available be used, but they also need to be available world wide.” And: “Validation remains an important
problem for all tools that might be used in a regulatory framework (Reed, 2008).”

To these I must add my own plea that suitable validated direct assessment tools be made available not just to navies, class societies, and
national maritime authorities, but to all practicing naval architects. This will ensure the widest acceptance and application, and be
rewarded with the most energetic independent research.

ADDITIONAL REFERENCES
CAGLAYAN, I. & STORCH, R., “Stability of Fishing Vessels with Water on Deck – a Review,” SNAME Journal of Ship Research,
June 1982.
ADEE, B.H., CHEN, F.I., Eberbardt, P., & Winandy, D., Fishing Vessel Dynamics and Stability, SNAME Spring Meeting/STAR
Symposium, Portland, OR, 1986.
PANTAZOPOULOS, M.S., “Three-Dimensional Sloshing of Water On Decks,” SNAME Marine Technology, October, 1988.
SHIN, Y.S., BELENKY, V.L., LIN, W.M., WEEMS, K.M., & ENGLE, A.H., “Nonlinear Time Domain Simulation Technology for
Seakeeping and Wave-Load Analysis for Modern Ship Design,” SNAME Transactions, 2003.
BELENKY, V., LIUT D., WEEMS K., & SHIN, Y., “Nonlinear Ship Roll Simulation with Water-On-Deck,” SNAME International
Ship Stability Workshop, 2002.
FALZARANO, J.M. & LARANJINHA, M., “Analysis of the Dynamical Behaviour of an Offshore Supply Vessel with Water on
Deck,” SNAME International Ship Stability Workshop, 2002.

The Second Generation of Intact Stability Criteria: An Overview of Development

LEE, Y.W., MCCUE, L., OBAR, M., & TROESCH, A., “Experimental and Numerical Investigation into the Effects of Initial
Conditions on a Three Degree of Freedom Capsize Model,” SNAME Journal of Ship Research, March 2006.
MACMASTER, A.G. & THOMPSON, J.M.T., “Wave Tank Testing and Capsizability of Hulls,” Proceedings of the Royal Society of
London, Series A, 446, pp. 217-232, 1994.

Prof. Lech Kobylinksi, Visitor

The paper provides an excellent overview of the development of the so-called second generation of intact stability criteria for all types
of ships, giving an explanation of why they have to be introduced and how they may be formulated as international recommendations.

The authors pointed out that modern ships cannot be assessed as safe solely on the basis of criteria included in the existing IMO IS
Code which were based on statistics of casualties that happened some forty or more years ago, and on regression analysis. The authors
mentioned also that risk analysis is now gaining acceptance as a tool for assessing safety levels in many complex industrial enterprises
and may be used in the development of new stability requirements.

The proposed three-level approach to the development of second generation stability criteria is a sound one and was already accepted
by the IMO SLF Subcommittee. This approach is based partially on risk analysis. In this approach, work concentrated initially on three
modes of stability failure, from which later on the dead ship condition was excluded for separate consideration. Those modes of failure
(or hazards) were recognized by the IMO Subcommittee in the mid-nineteen eighties, but because of lack of appropriate tools, this line
of development of criteria had not been followed.

Currently, in the development of second generation stability criteria, work was concentrated on restoring arm changes, parametric
resonance, and broaching-to. The main task was the assessment of vulnerability to those hazards. In the Level one and two criteria,
methods are proposed for this. However, the purpose of risk analysis is assessment of risk, risk being the product of probability of
hazard and its consequences. In the current work, the criteria for assessment of vulnerability as proposed are deterministic and in binary
form: yes or no, and not in terms of probability. Only in the third level, where direct assessment of stability is mentioned, the
probability will come out as the result of calculations. But direct stability assessment as mentioned by the authors could be a long-term
project and the discusser agrees to that. This is, however, a weak point of the current approach.

If the criteria were formulated in terms of probability, then the probability of meeting (or not meeting) the criteria could be used to
assess the safety level and to assess the need to apply some risk mitigation measures.

Generally the proposed approach may be considered as the designer approach. If, for example, in the design process, the ship, after
application of Level one and two criteria, appears to be prone to, say, parametric resonance, then what will happen? It is very
improbable that the body lines will be changed, because body lines are usually designed in order to achieve optimum propulsion
efficiency. Other means of mitigation of hazard probably will be used, including operational recommendations and possible
restrictions. However, at this stage it would be important to assess the probability of hazard and assess how those means of mitigation
reduce this probability. This is the point where experts’ opinions may be needed. Therefore, in the broader sense, experts’ opinions
should probably be included at some stages of assessment of the vulnerability criteria.

In spite of the critical comment above, the discusser believes that the currently presented approach is probably the best that can be done
at present. He strongly believes, however, that the future criteria should be formulated in probabilistic terms, taking into account
operational conditions and human factors using risk analysis methodology. It may be, however, a long way to achieve this goal. A goal
oriented approach to safety at sea is now considered as a proper approach to safety requirements by IMO.

The other point to which the discusser wants to draw attention is the choice of modes of failure. It is true that modes considered
currently are important. But the evidence of ship casualties due to stability failure shows, first of all, that all casualties result from a
sequence of events in which different factors play a part, the most important (about 80% of all cases) being the human factor. It is
important also to include in future considerations other modes of failure. For example, water on deck in extremely high and steep
waves (freak waves) caused several ships, including large bulk carriers, to perish with all hands without even MAYDAY. Water on
deck is a dangerous hazard also for some low-built fishing vessels. Shifting of cargo is another hazard with disastrous consequences.
Those are a few examples of other hazards that should be considered in the future, apart from the necessary revision of the existing
weather criterion.

The Second Generation of Intact Stability Criteria: An Overview of Development

Dr. Colin S. Moore, Member

Thank you to the authors for this comprehensive review of the ongoing developments in intact stability criteria. I appreciate the
emphasis on describing the physics of the phenomena we are trying to avoid. This is a valuable contribution. The paper reviews the
development of the two tiered vulnerability criteria approach. The purpose of the two tiered system is to be able to provide a quick
screening at the first level and then a more high fidelity assessment for ships that have been identified as being vulnerable by the first
level. It is surmised that some ships would be exonerated by the second level assessment, and that differences in vulnerability between
others identified.
The Level 1 criteria essentially encapsulate industry knowledge. Ship types that are considered vulnerable using these criteria are
expected to be vulnerable. This is effectively an application of experience in estimating the likelihood of stability failure in assigning
the criteria boundaries. The Level 2 criteria should be able to distinguish between ships identified by the Level 1 criterion.
This paper discusses various criteria and proposes some boundaries within those criteria, implicitly defining standards. For some ship
types, e.g. containerships, failure of these vulnerability criteria produces risks that are considered acceptable both by society at large
and the marine industry. Two steps are necessary, one to relate criteria to actual or comparative probability of failure in the mode being
tested, and two, to assign the acceptable risk. While definition of acceptable risk (i.e. the standards issue) is stated to be outside the
scope of these investigations, developing estimates of probability of failure or safety level is inside, and it is a key contribution to the
overall process.
For some criteria, vulnerability boundaries have been suggested based upon screening of results where there are large gaps between the
criterion values for ships in the sample populations that are known to not be vulnerable and those where adverse behavior is not
unexpected. However, the sample populations are small in size, and for some ship types, of relatively similar form. For convenience a
round number somewhere in the middle is chosen. However, no discussion of the effect of the boundary choice on the actual or
nominal probability of failure is included.
With respect to Pure Loss of Stability, on page 10 a step is taken towards developing a standard by comparing criterion values for
various ship types, some of which have be shown to be subject to pure loss of stability under certain conditions and others which
experience suggests are not, such as containerships. It is noted that the Level 1 criterion flags all tested containerships as vulnerable
and suggests that the Level 2 criteria should clear these ships. However, in the discussion of Level 2, no mention is made of how
containerships perform. The referenced SLF 53/INF.10 in fact does not clear the C-11 class containership (see Annex 2) under one
Level 2 criterion while clearing it in others for a specific sea state and speed (Annex 5). A summary or some preliminary conclusions
of applicability of the Level 2 criteria would be useful.
For Parametric Roll, both Level 1 and 2 criteria indicate that containerships are vulnerable to parametric roll. How does this impact
design and operation of containerships beyond what we already know? The key indicators at sea (i.e. the period ratio) for susceptible
ships and the appropriate mitigating behavior (i.e. change course) for large vessels have been identified. Further, the risk associated
with parametric roll seems to be acceptable for large vessels. For example, while there is public clamor over the perceived number of
containers floating in the ocean, industry appears to find its own, much smaller, estimate acceptable. Further reductions in these
numbers would be most reduced by improving or enforcing standards for and maintenance of cargo securing systems rather than
preventing rare rolling events.
Physics-based criteria have the advantage of being applicable to novel hull forms. Perhaps there are containership hull forms that are
not vulnerable to parametric rolling but still economically competitive. Developers of such hulls should investigate their behavior in
waves thoroughly. Would these criteria apply to a multi-hull, for instance? Application of criteria extended beyond their range of
applicability (mostly size) has been a problem for the industry.
I note the criteria are still in the development process and several alternatives are presented in mathematical detail. It is clear from the
discussion of the mathematics involved and computational tools required that evaluating some of the criteria will be burdensome. As
pointed out in the paper, current tools for direct assessment of dynamic stability remain to be validated. If the risk associated with a
mode of intact stability failure is small enough, then further pursuit of the criteria seems unjustified. It may be that there is insufficient
information to calibrate the probabilities of failure associated with the analysis approaches described for some modes of stability
failure. In that case it seems appropriate to concentrate on intact stability failures that have consequences that are unacceptable even if
of low probability.
This paper meets its objective to describe the current state of development of the second generation of IMO intact stability criteria. As
noted in the paper there is a clear distinction between a criterion and a standard. This separation is convenient for research but the
adoption of standards must be informed by an understanding of the strengths and weaknesses of the criteria development process, and
likewise criteria boundary proposals must include assessments of the safety level and the model sensitivity and uncertainty in their
choice. I encourage the authors to make these assessments as they proceed.

The Second Generation of Intact Stability Criteria: An Overview of Development

Prof. Marcelo Neves, Member
This paper highlights some interesting aspects of the complicated behavior of ships regarding such critical nonlinear phenomena as
pure loss of stability, parametric rolling, surf riding, and broaching. I would like to congratulate the authors for addressing these
potentially important instabilities and for writing such an interesting and readable paper about them.
The paper discusses these relevant problems within the perspective of development of the second generation of IMO intact stability
criteria. Given this specific perspective, whatever the case may be, it is evident that the theoretical models considered in this paper are
of a rather simple description of reality. I concur with the authors that this is a necessary approach, as general and practical guidelines
are the main objective; however, capturing the essentials of nonlinear phenomena but at the same time being simple is certainly a
problem of very complex nature. Therefore it must be recognized that the challenges for this qualified group of multinational experts
are huge. It is my understanding that the group is giving original and sound responses to these problems within the framework of
building up practical criteria, and I hope that the same team will continue their collaboration along these lines.
There are a few aspects of the paper which I would like to discuss with the aim of bringing my limited contribution to the
understanding of one of the topics of the paper:
1. Equation (30) aims to model some level of coupling between the vertical modes and the roll motion, corresponding to a hybrid 3-D
model. We have been examining the dynamics of parametric rolling trying to find answers to the following problems: a) Which
essential characteristics of global responses are lost when a simple coupling model is assumed? b) Are such losses comparable as
distinct hull forms are considered? Our results point to drastically different results for containerships as compared to fishing
vessels (Rodriguez & Neves, 2011). I would like to refer here to our results for a transom stern fishing vessel in head seas: when
compared to experiments the simple model gives excessive responses for a vast range of ship speeds, even in the range of
relatively small wave amplitudes. Unawareness of this aspect could lead to undesirable false positives when an uncoupled model is
assumed. This may be a relevant topic for future consideration as new stability criteria for different vessel types are still under
development.
2. Yet in both cases, either for the containership or the fishing vessel, the existence of upper frontiers for the domains of parametric
roll amplification was verified. This is an essential nonlinear aspect corresponding to strong internal excitation, as pointed out by
the authors in the introduction of the section on Level 2 Criteria for Parametric Roll. Yet it is not clear to what extent Equation (30)
would contemplate it, as the phenomenon depends on wave amplitude. The general well known result is the bending of the system
backbone to the right; see Figures 5~8 in (Neves & Rodriguez 2007). It is worth mentioning that this bending to the right (shift in
natural frequency) was observed in the last SAFEDOR benchmarking on PR as discussed in (Spanos & Papanikolaou 2009a,
2009b). It is our experience that this happens mainly due to the nonlinearities in the heave–roll–pitch coupling, as the nonlinearities
of the GZ curve become effective only after the roll motion is well developed. The contributions of the nonlinearities of the GZ
curve to the frequency shift may be to add or not to the tendency of the backbone to bend to the right, but the prevailing tendency
is given by the heave–roll–pitch coupling.
3. In (Neves & Rodriguez, 2007) we have analytically derived an expression for the shift in natural frequency (Equation (41) of
Neves & Rodriguez, 2007) produced by the heave–roll–pitch coupling which is proportional to wave amplitude squared.
Hydrostatics and wave passage (Froude-Krylov) effects are taken into account by means of integrals of hull geometrical
characteristics (longitudinal distributions of half breadth and hull flare) which describe the transfer of energy from the heave and
pitch modes to the internal excitation of roll motion. According to this, the frequency shift is a direct function of wave amplitude,
not of the resulting roll amplitude.
4. As this main frequency shift is of the type 2m4  kAw2 the constant k could well be examined as to be established as an alternative -
very simple – index for Level 1 vulnerability criteria, since large k means heavy tendency for parametric amplification. For
instance, k is very small for tankers but it tends to be large for e.g. containerships.
5. Finally, again regarding the frequency shift, it is important to mention that some ships (e.g. containerships) which are very prone to
parametric amplification may not respond at all at the e  20 tuning!

ADDITIONAL REFERENCES:
RODRIGUEZ C.A, & NEVES M.A.S. (2011) Domains of Parametric Roll Amplification for Different Hull Forms. In: Parametric
Resonance in Dynamical Systems, Fossen TI, Nijmeijer H (Eds.), Springer Science NY, Chapter 6. ISBN 978-1-4614-1042-3
NEVES, M.A.S. &, RODRIGUEZ, C.A. (2007) Influence of Nonlinearities on the Limits of Stability of Ships Rolling in Head Seas.
Ocean Eng 34: 1618 – 1630.
SPANOS D, & PAPANIKOLAOU A, (2009a) SAFEDOR International Benchmark Study on Numerical Simulation Methods for the
Prediction of Parametric Rolling of Ships in Waves. NTUA-SDL report, Revision 1.0, National Tech Univ of Athens.
SPANOS D. & PAPANIKOLAOU A (2009b) On the Decay and Disappearance of Parametric Roll of Ships in Steep Head Waves. In
Proc of the 10th Int Conf on Stab of Ships and Ocean Veh (STAB 2009). St. Petersburg: 259-270.

The Second Generation of Intact Stability Criteria: An Overview of Development

Dr. Arthur M. Reed, Fellow

The authors are to be thanked for a timely paper that brings the designer and owner/operator communities up-to-date with the direction
that the international standards community is heading with regard to intact stability. To this end some related experience of the U.S.
Navy may be of interest.

Since the early 2000’s, the U.S. Navy has had a performance-based dynamic-stability standard, (cf. ITTC, 2008; Reed, 2009 for
details). This standard was developed after it was found that the criteria of DDS 079-1 (U.S. Navy, 2007) did not provide the
equivalent margin against stability failure for tumble-home ship designs as it does for traditional wall-sided and flared designs. As is
the case of the 2008 IS code, the criteria of DDS 079-1 was based on the works of Rahola (1937) and Sarchin & Goldberg (1962) and it
did not sufficiently take into account unconventional vessels.

The U.S. Navy’s standard is, using the authors’ definition, a Second Generation criteria. As such, it has the disadvantage hinted at by
the authors of requiring significant computational efforts using a state-of-the-art computational tool that is capable of capturing all of
the physics of the dynamical system represented by a ship in extreme random seas. Thus, it would seem prudent to examine the
development of vulnerability criteria for naval vessels that would allow the separation of vessels that require the rigorous treatment of
the new U.S. Navy standard from those that do not.

An issue that is not raised by the authors, that needs to be considered, is the fact that the implementation of any of these
criteria/standards will be computerized, and thus must be Verified, Validated and Accredited (VV&A). As discussed in Reed (2008),
VV&A is a nontrivial process that cannot be taken lightly—a considerable amount of effort must be expended to perform this process
rigorously, and the costs can be significant, particularly if model tests are required to obtain validation data.

ADDITIONAL REFERENCES
ITTC (2008) The Specialist Committee on Stability in Waves: Final Report and Recommendations to the 25th ITTC. Proc. 25th ITTC,
Fukuoko, Japan, 36 p.
RAHOLA, J. (1939) The judging of the stability of ships and the determination of the minimum amount of stability especially
considering the vessel navigating Finnish waters. PhD Thesis, Technical University of Finland, Helsinki, viii+232 p.
REED, A. M. (2008) Discussion of: Belenky, V. L., J. O. de Kat & N. Umeda (2008a) Towards Performance-Based Criteria for Intact
Stability. Marine Tech., 45(2):122–123.
REED, A. M. (2009) A Naval Perspective on Ship Stability. Proc. 10th Int’l Conf. Stability of Ships & Ocean Vehicles (STAB ’09), St.
Petersburg, Russia, pp. 21–44.
SARCHIN, T. H. & L. L. GOLDBERG (1962) Stability and buoyancy criteria for U. S. Naval surface ships. Trans. SNAME, 72:418–
58.
U.S. NAVY (2007) Stability and buoyancy of U.S. Naval surface ships. Design Data Sheet DDS 079-1, Version 2.01, 30 January 2008.
NSWCCD Report NSWCCD-20-TR-2007/05, 107 p.

Prof. Zbigniew Szozda, Visitor

The development of the 2nd generation intact stability criteria is one of the main activities of the Sub-Committee on Stability and Load
Lines and on Fishing Vessels Safety (SLF), International Maritime Organization. The criteria used in the regulations contained in the
International Code on Intact Stability, 2008, are said to be the first generation. These regulations became mandatory after 1 st July 2010.
Nevertheless, the need for the development of the next generation criteria is very well justified. The paper contributes to this task and
may streamline further works. Therefore, the authors shall be congratulated for the paper containing the summary of the work already
done.

As this discussion should be limited in volume, I would like to make comments on three issues that are not directly related to the
physics and methodology used for the development of the criteria but which may facilitate further development.

DEFINITION OF THE STABILITY CRITERIA AND STABILITY REGULATIONS

I used to observe that people who are not directly involved in the discussions about the 2 nd generation intact stability criteria
experience difficulties in understanding the vulnerability concept – general approach, definitions, future application. Here I quote the
definitions of a criterion, standard, regulation and unconventional ship (SLF 51/WP.2):

The Second Generation of Intact Stability Criteria: An Overview of Development

 criterion is a procedure, an algorithm or a formula used for judgement on likelihood of failure;
 standard is a boundary separating acceptable and unacceptable likelihood of failure;
 regulation is a specification of a relationship between a standard and a value produced by a criterion;
 unconventional ships are ships that are vulnerable to stability failures neither explicitly nor properly covered by the existing
stability regulations.
At the current stage I totally agree with these definitions. But actually experts involved in stability matters have been using a word
“criterion” in a sense of “regulation” for a very long time. Once the definitions have been introduced all experts should demonstrate
very strong “discipline” in using these definitions both in writing and speaking. Especially, all the texts produced by SLF (as IMO
organ) shall be synonymous and precise. Additionally, new papers presented on workshops, conferences, new textbooks for students,
etc. shall use these definitions in the same manner. It also puts a new light on the current IS Code, 2008 edition. For example, the
sentence used in paragraph 2.1.7 – “If curves or tables .... are used to ensure compliance with the relevant intact stability criteria,....”
does not express proper meaning of the word “criterion”. Therefore, in case of revision of the current IS Code, before issuing a new
edition, the sentences containing the words “criteria, standard, regulation” shall be revised and amended appropriately. This is one of
future tasks for SLF emanating from current work.

MULTI-TIERED APPROACH
A multi-tiered approach to the 2nd generation intact stability criteria has been agreed so far. A proposal containing the flow chart of the
approach was submitted to SLF 53 in the document SLF 53\3\5. Three tiers were proposed there. As the result of the discussion at SLF
53, this flow chart was changed. A temporary version is shown in the paper in Figure 1. The consequences are considerable (the fourth
tier was added) and in my opinion this change does not reflect the spirit of the “safety level”.
Entry to Direct Assessment means that the ship is vulnerable to particular stability failure mode. It has been checked on the Level 1
(using simple criterion) and Level 2 (using more complex, sophisticated method). To my understanding, exit “No” from Direct
Assessment means that the Administration is not satisfied with the results of this assessment – the required level of safety has not been
achieved. I doubt whether such a ship may be designed, constructed, and put into operation even if an Operational Guidance (fourth
tier) has been issued for the Master. The aim of the Operational Guidance is to assist the Master of a particular ship in avoiding
dangerous situations and to prevent the ship (designed on acceptable level of safety) from possible stability failure. In my opinion this
part of the flow chart needs further discussion at the next SLF meeting, taking into account the knowledge that a considerable
percentage of stability accidents were caused by human factors.
RELATION BETWEEN LEVEL 1 AND LEVEL 2 CRITERIA
Multi-tiered approach will serve the shipbuilding industry effectively in terms of safety level if there is proper relation between the
Level 2 and the Level 1 criteria. The authors mentioned in the paper that “false positives” on Level 1 may be expected. It is a
consequence of the assumptions to the approach. I would like to name here three rules concerning relations between the Level 1 and the
Level 2 criteria that should be met in practice.
The first rule is extremely importants. A ship that has been found vulnerable on Level 2 (which is more precise than Level 1) has to be
found vulnerable on Level 1 as well. Otherwise there might be a possibility that ships actually vulnerable to a particular stability failure
mode may not be found on Level 1.
The second rule is also important. A ship that has been found non-vulnerable on Level 1 has to be found also non-vulnerable on Level
2.
Additionally, the third rule – ships which suffer from particular stability failure mode in service shall be found vulnerable both on Level
1 and Level 2.
The question is how to prove these rules before making practical use of particular criteria. I have no answer to this question at the
moment. In my opinion it has to be proved rather by mathematical considerations than by calculations performed on a limited sample of
ships. But of course the criteria are under development and the discussions during next SLF sessions and other stability conferences and
workshops obviously will put new light on this issue.
CONCLUSION
There has been very active development of the 2 nd generation intact stability criteria based on vulnerability concepts over the past
several years. The authors’ personal and respective institutions’ contribution to the task is magnificent. The paper summarises the state-
of-art and gives examples of possible new criteria and standards. Nevertheless, there are still a number of scientific and practical
problems which have to be solved before the 2 nd generation intact stability criteria could be applied to ship design. It requires a lot of
future effort and international co-operation.

The Second Generation of Intact Stability Criteria: An Overview of Development

ADDITIONAL REFERENCES
IMO 2008, International Code on Intact Stability, 2008, adopted by Resolution MSC.267(85) on 4 th December 2008, London.
IMO SLF 53/3/5, Comments on the structure of new generation intact stability criteria, submitted by Poland, London, 2011,
IMO SLF 51/WP.2, Report of the Working Group (part one), London, 2008.

Mr. John Womack, Member, and Prof. Bruce Johnson, Fellow

Our first comment concerns the paper’s basic assumption in the section “Changing Stability in Waves” that only the inertia of the
waterplane area in the fore and aft sections has an effect on dynamic stability is greatly oversimplified for any vessel not having a
parallel middlebody and a midships section coefficient much less than one. Any deviation from this simple hull shape (used by some
commercial ships) makes the basic assumptions no longer valid. And since most small vessels have little to no parallel middlebody nor
are walled side this assumption and the criteria developed from it will be invalid for the historically most vulnerable class of vessel’s
afloat.

The following approximation is useful for far more vessels than the authors’ analysis as it captures the changing hull shape over the
depth of the hull:

GM  KB  BM  KG
T  5 CB  CIT B 2
     KG
3  2 CWP  CB 12T
Note, however, that CWP, CB, and CIT frequently vary with draft, except for the special cases of rectangular barges and completely wall-
sided ships. So the quoted statement from the paper is greatly oversimplified except for these special cases. The localized draft still has
a very large effect on a sectionalized GM when “perched on a wave”.

Further, the section “Changing Stability in Waves” needs to have a very large caveat stating that it is only valid for vessels with
freeboards that exceed at least half the wave heights being experienced. If the wave crest at amidships is higher than the freeboard the
stability, GM and RAs, will be greatly reduced, both from the dramatic loss of waterplane area and from the water on deck. It was
these waves that created the negative GMs and negative RA curves from loss of waterplane only in the A Systematic Study of Wave
Phasing on Righting Arm Curves for Fishing Vessels by John Womack and Bruce Johnson (SNAME Transactions, 2005, not
referenced in this paper). Waves that were not greater than the amidship’s freeboard only caused modest to small reductions in both the
GM and RAs.

For example a 100 foot fishing vessel will often have freeboards in the 4 foot range. A 100 foot long wave with a wave height of only
10 feet will have the wave crest flooding the deck at amidships which will dramatically lower GM and RAs. This is a wave with a
length to height ratio of 10, steep but not breaking, that will be likely be experienced many times during a vessel’s life.

Figure D-2: Guideline of critical speed

The Second Generation of Intact Stability Criteria: An Overview of Development

The condition is even worse for waves up to 2.5 times the length of the vessel which could have wave heights of 25 feet for the same
1/10 slope (See Figure 16 from Kan 1990). The following quotes from Kan (1994) bear repeating, “It has been said that the mo st
dangerous wave length is λ/LWL = 1. However, it is obvious from these tables that many capsizings were observed in longer waves such
as λ/LWL = 1.25 to 1.75” (Note: beyond which the steepness decreased to 1/20 in these tests because of wavemaker limitations.)
“Therefore, we should recognize that the most dangerous wave length is not restricted to λ/LWL = 1…..”. Thus dramatic drops in GM
and righting arms can occur at wave lengths much longer than previously assumed worst case condition of equal to boat length.
Based on this there is no need to check for the vulnerability for pure loss of stability in low freeboard boats. They are theoretically
vulnerable, just move directly to the criteria phase (small boats, big waves, common sense). Small boats for the purpose of pure loss of
stability can be easily defined as those whose freeboard is equal to or less than the height of the steepest waves that will be experienced.
This leads to the question why develop new criteria for pure loss of stability (riding on a wave crest). It is only a potential risk for
small boats and to our knowledge, no vessel that met currently established stability criteria in recent history has been lost from riding
on a wave crest without also taking on water either trapped on deck or starting the downflooding sequence. This condition represents
impaired stability rather than intact stability, which the proposed criteria do not address.
This has been shown in model studies of small fishing vessels in irregular waves (Grochowalski, 1989, 1990, 1993, 1994, but not
referenced in this SNAME paper) that showed that those vessels that capsized were stability deficient to start with and then succumbed
only after a unique series of wave encounters that first put large quantities of water on deck that severely reduced stability and was then
followed by a larger than average wave. If a smaller than average wave was next the vessel generally had time to clear her decks and
would survive the encounter.
For a boat to experience the pure loss of stability the criteria attempts to address the operator would need to be running the boat at very
near the average wave speed of the seas for a long period of time to catch the rare wave group that so closely matched the boat speed
for a significant period of time and was steep enough to exceed the vessel’s freeboard. This sequence of wave events is a rare
occurrence in actual seas which, unlike regular wave tests in a towing tank or numerical simulations, are just too irregular for a boat to
remain on a wave crest for any length of time to roll over, even with a negative GM, thus the reduction in GM and RA passes
harmlessly. Further, when the operator nears the speed of the wave groups the boat starts to feel “loose” as a wave slowly passes by
and the operator will (or should) slow down to allow the waves to pass by quickly.
An additional problem is the complexity required to develop the proposed level 1 vulnerability criteria coefficients C CWS and the
proposed level 2 vulnerability criteria value “time-below-critical-GM”. For small boats as discussed above, the proposed 2 nd
generation criteria add unnecessary mathematical complexity to an unrealistic physical model. This results in an unneeded significant
level of additional work that will only increase costs with little chance of improving safety, especially in the smaller vessel area.
If one though wishes to have a criteria for small boats as defined above just calculate the actual GMs and RA curves with the vessel
perched on waves the vessel is expected to see. GMs and RA curves are already calculated in the stillwater condition so the hull model
and loading conditions are already in the software data. For all of the current basic stability analysis software used for small boats such
as SHCP, GHS, and Hydromax, putting the vessel on a wave and running GM and RAs is just a couple of keystrokes. This would be a
small amount of additional work and would give the designer and operator useful knowledge about the vessel.
One could then develop minimum standards for allowable reduction in GM and RA curves from current still water criteria levels. As
noted above though, we know of no vessels lost because of pure loss of stability without shipping water on deck so why bother. The
current stability criteria that exists works and is relatively easy to calculate. The proposed 2 nd generation criteria only adds unnecessary
complexity and cost to regulations which have little chance of improving SOLAS, especially in the smaller vessel area.
The same comments as for pure loss of stability apply to the criteria development for Surf-Riding and Broaching. Small boats will also
be vulnerable to broaching pretty much as a fact of life. Once again skip the check for the vulnerability and just move directly to the
criteria phase (small boats, big waves, common sense). Again, the proposed 2 nd generation Intact Stability criteria appears to add
unnecessary mathematical complexity to an oversimplified physical model (including regarding surf riding as a stationary condition)
resulting in an unneeded significant level of additional computational overhead. This proposal will only increase costs with little chance
of improving safety, especially in the smaller vessel area. The best “criteria” to date and always will be is operator training not to run
the boat near wave speed.
In summary, the current internationally recognized stability criteria; area under righting arm curves, water on deck, and severe wind
and roll; have generally worked well for vessels over the accepted 79 foot cutoff. There is no need to reinvent the wheel. For vessels
under 79 feet all that is needed is modification of the basic criteria’s values is in order to scale them to the smaller vessels in smaller
scaled wave conditions (No “one size fits all” waves). This is the best approach as the basic criteria are well known, are easy to
calculate, easy to explain to the vessel operators, and have worked when common sense about avoiding extreme wind and waves is
observed!

The Second Generation of Intact Stability Criteria: An Overview of Development

 Authors’ Closure
We want to thank each of the discussers of our paper for their insights and contribution to this subject.

Dr. Corrignan’s inquiry regarding the generalization of using a probabilistic approach to all phenomena is timely. At IMO’s Maritime
Safety Committee, the development of the “Safety Level Approach” that incorporate probabilistic principles is under active
consideration and is promoted by its advocates as applicable to many areas of maritime safety. With respect to stability, we agree with
the discusser that the probabilistic approach is desirable; it provides consistency throughout the process as the estimate of probability is
a universal measure; the “overarching” probabilistic approach is possible if we will be able to demonstrate that the mathematical
models are compatible, i.e. probabilities estimated with different models can be favorably compared to each over:

Surf-riding (broaching) is one of those “unconventional” problems for as it combines hydrodynamics and nonlinear dynamics. We
agree that more explanations are needed on the phenomenon and its mathematical description. We also agree that broaching is a new
“animal” for approval and requires data on powering that is often confidential; at the same time, regulatory bodies deal with hull forms
that are also confidential information.

Information in operational guidance is mostly useful when an operator is fully aware of environmental conditions. At the same time,
on-board wave-measuring technology has not yet matured; therefore the guidance should be useful without such a device. While most
large (and fast) ships are weather routed; the information available through the weather routing services is complete enough to make the
operational guidance useful. Another benefit is that crew becomes aware of increased risk of stability failure up to 6 hours in advance,
which will help an operator to make a more informed decision in consideration of the ship-specific guidance.

In response to Dr. de Kat, we observe different approaches are being considered for the selection of reference waves for vulnerability
criteria levels 1 and 2 for the two failure modes involving righting lever variations in waves, namely pure loss of stability and
parametric roll. One of these involves concurrent consideration of two methods for the wave conditions; for one, a sea state is presented
as a set of regular waves with random length and height; the second round of averaging is done for a sea state using the wave scatter
diagram. A joint distribution of wave heights and lengths accounts for their dependence. The second approach uses regular waves that
can be considered as “equivalent”. In this case appropriate corrections may be necessary to account for a fact that longer waves are
rarer. Another view considers the worst wave steepness in association with the hull geometry that includes consideration of a wave
steepness estimate using Grim’s effective wave concept.

The level 1 vulnerability criterion for surf-riding takes into account ship length and reflects the fact that steep waves capable of causing
surf-riding have an encounter likelihood that decreases as the ship length increases. The level 2 criterion reflects more individual
features of a ship, since the calculation includes an assessment of the surging force as well as an approximation of actual thrust and
resistance.

At the time of this writing, the standards for the criteria were not considered, so a roll angle of 20 degrees was used for the purpose of
the numerical example only. We agree with the discusser that the both levels of vulnerability criteria should be sensitive to loading and
environmental conditions that are reasonably expected in operation. With this in mind, we advocated a level 1 criterion for parametric
roll with an option to account for different wave cases as well as a realistic range of ship service speeds.

We are very appreciative of Prof. Francescutto’s recollection of Pierrottet’s response to Sir Archibald Denny and how it is apropros to
the project at hand. Our attention on simplicity at the level 1 criteria is focused at the design naval architect, who should only have to
consider the detailed assessment of vulnerability addressed in the level 2 criteria for a vessel design and loading condition for which the
level 1 can assess a potential vulnerability. If this can be reliably achieved using simple approaches, naval architects who have the
responsibility of showing compliance and of verifying that compliance are sure to appreciate the effort involved.

Mr. Hutchison observed that the paper mainly addresses the three stability failure modes of pure loss of stability, parametric rolling
(both of which are associated with righting lever variations in waves), and broaching-to. The identification of water on deck as well as
deck in water as but two other stability failure mechanisms for which criteria may be considered for development in the future. The
chief reason that the work has concentrated on the failure modes mentioned is that they were specifically identified in the 2008 Intact
Stability Code.

While the work identifies surf-riding as a pre-requisite for broaching-to, Mr. Hutchison correctly identifies that some vessels may be
subject to broaching-to without surf-riding. However, the assessment of maneuvering capabilities becomes an essential element for any
performance assessment of a ship’s vulnerability to broaching without surf-riding. Because the development of such an assessment
has, at this time, not been initiated, our paper does not address broaching-to without surf-riding, but would agree that this phenomenon
could be added to a future development.

The Second Generation of Intact Stability Criteria: An Overview of Development

We appreciate the added perspective that Mr. Hutchison provides with respect to tugboats when operating with and without a towline
force and agree that this consideration should be included in the guidance that should be provided to tugboat operators.

As regards equivalent linear damping in Equation (33), the cautions the discusser presents are well appreciated. However, because the
criteria strives to identify the onset of parametric roll, the shortcomings associated with equivalent linear damping to the large roll angle
are not so critical as that associated with smaller roll angles.

The concern about an excessive level of conservatism associated with the application of Melnikov’s method to surf-riding has not been
exhibited in the calculations of sample ships. This suggests that the criterion formulation is consistent and reliable.

We welcome Mr. Hutchison’s interest in the problem of rarity as it manifest in ship dynamic stability issues and we share the
observation of the strong non-linearity of ship dynamic stability. Clearly, to properly address this issue requires a balanced approach
that is also feasible.

While we might want to join the discusser to advocate for the broad availability of validated direct assessment tools, the possibility of
achieving such a laudable goal – in our view - requires overcoming obstacles such as proprietary rights to direct assessment tools that
are the result of substantial resource investment. Innovative mechanisms to address these obstacles require further inquiry,
development, and – surely – negotiation.

We thank Prof. Kobylinski for his insightful comments. Development of direct assessment tools and procedures is, indeed, a
challenging undertaking. The existing experience with advanced codes and their application is most relevant to advance design
analysis. The main challenge is how to convert this experience for the regulatory use.

Human element has to be considered when the decision is made: to change the design or to develop an operational guidance and how
difficult it will be for the human operator to follow these guidance. Expert opinion here would be quite appropriate.

In our view, one of the reasons why SLF has chosen these particular modes of stability failure is that existing knowledge and
technology is mature enough to address these modes in a regulatory framework.

We agree with the discusser that the probabilistic approach is the way forward as it is the natural tool to describe the intrinsically
random environment in which a ship operates.

As it is correctly noted by Dr. Moore, the paper describes the current state of development of the vulnerability criteria and this is the
reason why some of the issues were not addressed in the paper. However, as the further development proceeds, some of the answers
become available.

Modern container carriers generally are not known to be susceptible to pure loss of stability failure. However, large variation of the
righting lever in waves causes the level 1 criterion to identify them as potentially “vulnerable”. The result of the second level
vulnerability analysis was not yet clear when this paper was written. Further development now shows clearly that use of GM for the
level 2 should be substituted by another stability parameter, perhaps range of positive stability. By making this substitution, results may
be obtained that do point out ships that are known to have pure loss of stability vulnerability and do not indicate such vulnerability to
other ships.

We agree with the discusser that susceptibility of modern containerships to parametric roll is well known in the maritime community.
Ability of the level 1 and 2 vulnerability criteria to indicate this susceptibility is a good sign that the criteria properly account for the
essential physical properties of the phenomenon. Since only hull form and loading conditions are used for inputs, we anticipate that the
criteria will correctly identify susceptibility to parametric roll for a new design. We also believe that the criteria can be applied to
multi-hull ships.

As far as the sample ship population is concerned, many different ships (now about 150) were used for calculations since the paper was
written. The results are expected to be available at the 55 th session of SLF (February 2013).

The discusser’s views on the potential for evaluation of some of the criteria to become burdensome have been included in the ongoing
discussions in the development of the criteria. Clearly, the tasks of demonstrating and verifying compliance with criteria should be
within the technical capabilities of those who are assigned these tasks. We agree with the discusser that a focus on stability failures that
have unacceptable consequences is appropriate even if the failure likelihood is low. Assessment of safety level for identifying the
proper standard will remain a key element of the development.

The Second Generation of Intact Stability Criteria: An Overview of Development

Hull geometry (especially) stern shape has a significant influence on the development and severity of parametric roll, which was shown
by Prof. Neves in the referred work, by research group at MARIN (Levadou et al 2006), and others. The motivation for a 3D hybrid
model was to include most of ship-specific features in the simplest analytical model possible. This hybrid model includes full
hydrostatic coupling between heave and pitch; coupling with roll is one-sided, which may be a source of conservatism of the model
because the transfer of energy from roll to other degrees of freedom is not modeled.

The model (Equation 30) is currently under validation. We agree with the discusser that these nonlinear phenomena (frequency shift,
appearance of upper boundary) should be tested during the validation so that the need for the missed coupling for the vulnerability
check can be verified.

We agree with the discusser that the worst case of parametric roll may be different from exact linear conditions. This effect, however,
may be caused by significant initial curvature of the GZ curve, which is the case for some high free-board vessels, containerships
included. That is the reason why the second vulnerability checks need to be performed for a series of speeds, i.e. for the series of
encounter frequencies.

We agree with Dr. Reed that the performance-based criteria that are needed for unconventional vessels may require significant
calculation and, therefore, incurred additional expenses. Vulnerability criteria are seen as the way to justify these expenses by
identification of the designs that may benefit from the direct stability assessment.

The paper is focused on vulnerability criteria as they are the current subject of interest in the development of the second generation
IMO intact stability criteria. The consideration of problems related to direct stability assessment has been just started. Nevertheless, it is
clear that the issue of validation is the central one for application of numerical simulation for regulatory purposes.

We agree with Dr. Szozda that the development of the second generation of intact stability criteria is a large technological leap
forward. Clear and strict use of terminology is very important, not only for “discipline of thinking”, but as a means of coordinating the
activities of experts from different countries; once agreement is reached on criteria, the development can proceed towards standards.

It is not yet clear how the results from the direct assessment will be used in design and regulatory practice. One possibility, if the
Administration is convinced that it is not possible to design a safer ship for the intended application, the restricted operation can still be
allowed with the appropriate operational guidance that provides details of those restrictions. In this case, at least, all parties involved
are aware of the high risk associated with this ship. It is important that such a decision is made “with all eyes open”.

The question on “three rules” of multi-tiered criteria is directly related to validation of those criteria. The numerical example, included
in the paper is not yet a complete validation, but rather a demonstration that the proposed ideas for the criteria make sense. Complete
validation of the criteria seems to be possible only after direct assessment procedures are completed. Then it will be possible to
compare safety levels and demonstrate that lower tier criteria are, indeed, more conservative.

We appreciate the interest and concern that Prof. Johnson and Mr. Womack have for this subject and their discussion shows that we
were remiss in not referencing the contributions made by their 2005 SNAME Transactions paper to the assessment of ship stability in
waves. That paper summarized results from the SNAME T&R project which investigated the effects of longitudinal waves on the
intact righting lever characteristics of smaller fishing vessels. Inasmuch as the development work described in our paper is addressed at
passenger and cargo ships of 24 meters in length and greater, small fishing vessels will be outside the scope of this development. That
said, the key aspects of their 2005 paper that the discussers point to are elements that have been taken into account in the current stage
of development.

For example, concerns about the waterplane area have now been incorporated by taking into account a check for potential vulnerability
that includes the metacentric radius (BM) as well as stability when the ship is perched on a wave. We also agree with the observation
of the reductions in stability that would occur where a wave crest is higher than the freeboard. The need to evaluate stability in
wavelengths that vary to values greater than the ship length is understood and is well illustrated in the discusser’s 2005 paper.
However, the development to date has considered using the wavelength on the order of the ship length to identify ships for which a
potential vulnerability to pure loss of stability exists and which would then require more thorough evaluation of the ship’s stability in a
series of waves of different sizes. This process has been pursued for a number of sample ships that include at least two ships that have
been identified as vulnerable to the pure loss of stability failure mode.

As regards surf-riding and broaching-to, we agree that the first level check should be very simple and for this reason the current view is
to proceed to a thorough assessment of vulnerability if both the ship size (i.e., length) is less than a certain threshold and the Froude
number is greater than 0.3. We understand that operator guidance to avoid surf-riding conditions, such as that contained in IMO
MSC/Circ. 1228, will always provide a good means to keep the ship safe – if it is followed.

The Second Generation of Intact Stability Criteria: An Overview of Development