FAILURE ANALYSIS OF HIGH-PERFORMANCE SURFACES USED FOR TRANSVERSAL STABILITY

OF SHIPS (BILGE KEELS)

1, a
Hugo Rodrigues, 1, b, 3 Rui F. Martins, 2, c L. Leal das Neves, 2, d Paulo Silva
1
Faculdade de Ciências e Tecnologia, Universidade Nova de Lisboa
UNIDEMI, Departamento de Engenharia Mecânica e Industrial
Campus de Caparica, 2829-516 Caparica, Portugal
Email:a [email protected], b [email protected]
2
Portuguese Navy, Directorate of Ships, Studies Department
Naval Base of Lisbon, 2800-001 Alfeite, Portugal
Email: c [email protected] , d [email protected]
3
ICEMS/IST, Av. Rovisco Pais, 1, 1049-001 Lisbon - Portugal

ABSTRACT

During a scheduled maintenance program of a ship, several cracks were detected in the internal structure of bilge keels,
as well as on the connection of these to the hull. The cracks encountered compromised the watertight integrity of the
hull and, in the long run, could jeopardize its integrity.
This paper presents a failure analysis of the structural damages detected, in order to identify the causes of failure and
propose improvements that could avoid them. Several stress concentration areas were identified, which coincided with
the critical areas where there have been found the structural failures. It was confirmed the reinforcements (stiffeners)
influence in the increased levels of stress in the welded joints of the bilge keels to the hull.
Alternative design geometries for the bilge keel internal structure are presented as well as it is shown that the
application of brackets or a bulb flat on the connection of the bilge keels to the hull can reduce the high level of tensile
stresses that are induced in the structure.
The fatigue analysis according to the Germanischer Lloyd Classification Society is applied to the new structural
internal geometry and is presented on this paper.

KEY WORDS: Bilge Keels, Structural Failures, Failure Analysis, Improved Design Arrangement, Finite Element
Analysis.

1. INTRODUCTION condition, possesses good weldability and low harden

effect after welding, due to the low probability of
The hydrodynamic resistance of a ship to roll motion martensitic transformation during cooling [2].
can be increased through passive and/or active
stabilizers. Encased in the passive roll stabilizers are the
bilge keels under study, which, in addition to attenuate
the roll motion, present a low manufacturing cost,
compared with the cost associated with the manufacture
of active stability systems.

In essence, bilge keels are hull appendices, applied

approximately at half length of the ship, on both sides
of the hull, which transform the vessel’s motion kinetic
energy into fluid motion, especially under the form of
vortex shedding [1]. Their efficiency depends on their
location, size and shape [1]. The bilge keels studied and
presented in this article have a triangular, or “V”,
arrangement as shown in Figure 1.

The bilge keel in study is built in D36 steel,

characterized by the chemical composition and
mechanical properties referred in Table 1 and 2 [2], Figure 1. Cross section view of the bilge keel with a
respectively. The steel, furnished in the normalized “V” arrangement studied. Examples of cracks detected.
.Table 1. Chemical composition of D36 naval steel watertightness and the separation of the welded
grade type [2]. connection/joint between the two plates that define the
C [%] Si[%] Mn [%] P [%] S [%] Al [%] external surface used for the transversal stability of
0.18 0.5 ≥0.7 0.035 0.035 ≥0.02 ships (Fig.1).

Table 2. Mechanical Properties of D36 naval steel 2.4. Structural Failures Location
grade type used in the bilge keels manufacture [2].
Yield Strength Tensile Strength Elongation [%]
Both the cracks in the bilge keel internal reinforcements
[MPa] [MPa] and the cracks at the plating edge occurred along the
≥355 490-620 ≥21 entire length of the bilge keel.
The remaining cracks (Figure 3, points marked) were
The connection to the hull is performed through a located in the middle plane of the internal
doubler plate (Figure 2), where a full welding reinforcements of the hull (bulkheads /frames) and, in
penetration without any intermittent fillet welds is some cases, also in the alignment with the middle plane
imposed. The doubler plate connection to the hull is of the bilge keel internal reinforcements (Figure 3).
often encountered in marine structures, not only due to
its easiness in production, because it provides a faster
and easier way to perform the alignment of the bilge
keel over the hull structure, but also represents an
excellent configuration to avoid multi-axial stress states
at the connection to the hull [1].

Figure 3. External overview of the bilge keel. Recorded

location of cracks (marked points).

3. DESIGN LOAD, PBK

Figure 2. Doubler connection detail of the bilge keel to While one can often adequately predict heave, pitch,
the hull. sway, yaw and even surge, roll motions of a ship are
still remarkably difficult to predict, since roll is the
In case of an accident or catastrophic failure, the bilge motion that is very dependent upon viscous effects of
keel will partly, or entirely, detach from the ship's hull the fluid on the surfaces used for transversal stability of
structure without causing structural damages that could a ship [3]. Nevertheless, design principles from the
compromise the hull plating water tightness. Germanischer Lloyd Classification Society were used
[4], in order to determine the maximum load applied on
the bilge keels, PBK (fig.4) (1)
2. FAILURE REGISTER

2.1. Cracks in the Bilge Keel Connection to Hull

Several cracks were detected by visual and dye

penetrant inspection up to 150mm long, mainly in the
welded connection between the bilge keel plating and
the doubler plate (Fig.1), but also in much smaller
number and size, on the doubler connection to the
ships’ hull plating. .
Figure 4. Design load, uniformly distributed across the
2.2. Fractures in the Internal Structure of Bilge Keel bilge keel span.

After a close inspection at the internal structure of the Equation 1 is valid for ships with a length L between
bilge keel, multiple fractures, originated from fatigue 50m and 200 m, measured in the corresponding
cracks, were recorded. These occurred at the waterline and for bilge keels located between 0.4L and
intersection between the longitudinal and transverse 0.6L.
reinforcements of the bilge keels, leading to the
PBK =
52000 × d
[
× γ fdyn kN / m 2 ] (1)
complete separation of the link between them (Fig.1).
(L + 240 )1.1
2.3. Fractures in the Bilge Keels Plating Edge
Where, d, represents the density fluid value, involving
At the bilge keels plating edge multiple fractures were the bilge keels (of 1.025, for saltwater at 25ºC), L the
recorded, resulting in the loss of the bilge keel distance measured from bow to stern, on the waterline
(109m, for the case in study), and γ fdyn the partial When submitted to an uniaxial tensile test, some
mechanical properties of the D36 steel used in the bilge
safety factor representative of permanent and cyclic keel construction (Table 1 and 2) are characterised by
loads acting on the undamaged structure in normal its true Stress–Strain curve, as shown in Figure 7 [5].
operational conditions (considered a value of 2). The calculated coefficient and exponent of the
Accordingly, a design load ( PBK ) of approximately Ramberg-Osgood Law, K and n, were 879.2 MPa and
170kPa, was calculated. 0.1634, respectively, and the stress-strain curve was
considered in the definition of the FE material model.
Since the waterline length of the ship, where the bilge
keels under study are installed, is 109m, the domain of
validity for the design load is satisfied; regarding the
second parameter on the range of validity of the
equation, between 0.4L and 0.6L, the design load is
only valid for approximately 70% of the length of the
bilge keels under study, as shown in Figure 5. Although
the region where the application of the design load is
valid does not include the entire length of the bilge keel,
it covers some of the registered structural failures (Fig.
5), not invalidating the calculated value of the design
load and its implementation in subsequent studies.
Figure 7. True Stress-Strain curve for the D36 naval
steel [5].

The regions identified as hot spots were numbered

Figure 5. Valid domain for the design load, overlapped (Figure 8), having been registered their corresponding
with the location of the recorded cracks. Von Mises stress values (Table 3).

4. BILGE KEEL GEOMETRY MODELING AND

ANALYSIS BY FINITE ELEMENT METHOD
(FEM)

4.1. Nonlinear FEM Analysis of a Bilge Keel Section

A bilge keel section with 200 mm long was modeled, in

order to perform an analysis of the induced stress and
deflection on the structure due to the design load. The
model shown in Figure 6 includes the internal structural
arrangement of the bilge keel and a hull section of
400x500mm length with a 3m curvature radius. Figure 8. Bilge keel hot spots location.

The three-dimensional model was uniformly loaded Table 3. Maximum Von Mises stress values located in
with 170kPa in the upper face of the bilge keel plating, the hot spots for the nonlinear Finite Element Analysis
corresponding to the design load (PBK). The edges of the (FEA) performed.
hull were fixed, in order to simulate the boundary
conditions present in the real structure (Figure 6). Total of Total of Hot Hot Hot Hot Hot
Nodes elements Spot 1 Spot 2 Spot 3 Spot 4 Spot 5
229 317 336 364 359
50680 24702
MPa MPa MPa MPa MPa

The data presented in Table 3 shows high levels of

stress concentration located in the hot spots, inducing
local plastic deformations, but also yield strength safety
factors close to one, which leads to the nucleation and
propagation of low-cyclic fatigue cracks.
Figure 6. Three-dimensional model of the bilge keel
section modeled, which includes the internal structural With reference to the yield strength value of the
arrangement, the finite element mesh, the design load structural steel in which the bilge keel is made of
and the boundary conditions. (355MPa), the hot spots numbers 4 and 5 entered the
plastic regime, concurring with the location of the
register structural failures (Figure 1).
4.2. Study of the Influence of the Hull Reinforcements’ details is the application of a steel bulb with 15mm in
Alignment with the Internal Structure of the Bilge Keel diameter in the bilge keel plating edge (Figure 10).

Having been identified as hot spot 1, the stress Table 4. Alternative internal geometries
concentration at the bilge keel to the hull connection Reference Description  Illustration 
(Figure 8), and having been detected a pattern in the
location of the structural failures, namely the Model containing only
registration of cracks in the alignment of the hull one solid and continuous
Geometry transversal stiffener.
internal reinforcements, an analysis of the position nr: 1
influence of the bulkhead/frames in the induced stress
values in the connection between the bilge keel and the
hull was carried out. Model containing one
transversal and
Geometry longitudinal stiffener,
As shown in Figure 9, it was found that the presence of with smooth cuts in the
nr: 2
the hull’s internal reinforcements (bulkheads or frames) intersection between
resulted in an increase of the maximum stress value them.
present in the connection between the bilge keel and the Model containing only
hull, since the maximum induced stress of 230 MPa one transversal stiffener
increased to approximately 260 MPa (+13%). In Geometry with two sections, with
soft cut contours.
addition, there was also a maximum stress concentration nr: 3
in the region closest to the alignment of the
reinforcements. Model containing one
transversal and
Geometry longitudinal stiffener,
nr: 4 with smooth cuts in the
intersection between
them and with two
sections with soft cut
contours
Model containing only
one transversal stiffener,
Geometry with multiple circular
nr: 5 cuts sections and stress
concentration reliever
holes.

Model containing one

Figure 9. Influence of the alignment between the bilge transversal and
keel and the hull transversal reinforcement. longitudinal stiffener,
with smooth cuts in the
Geometry intersection between
nr: 6 them, and with multiple
5. ALTERNATIVE GEOMETRIES circular cuts sections
and stress concentration
relievers holes.
Taking into account the results from the FEM analyses,
new geometries were designed and analysed in order to
propose a viable substitute for the bilge keel internal
structure and its connection to the hull. Detail A

5.1. New Internal Bilge Keel Geometry

With the objective of maintaining unchanged, as much

as possible, the hydrodynamic characteristics of the
bilge keel, in particular those related to its energy
dissipation and drag coefficient, only the internal
geometry of the bilge keel was redesigned, maintaining A
the thicknesses of the components, so that the rigidity of
the structure was kept unchanged. The new developed Figure 10. Stress concentration relievers.
internal geometries are shown in Table 4.
All the tridimensional models of the new bilge keel
Several details were included in the new internal internal arrangements were uniformly loaded with
arrangements in order to minimize the stress 170kPa, corresponding to the design load. Also the
concentrations (Figure 9) and, concomitantly, facilitate same boundary conditions as shown in Fig.6 were
its production. An example representative of these applied, in order to compare the results with those
obtained for the current model.
Table 5 presents a summary of the results obtained from the bilge keel connection (Figure 11b). A two-
the FEA, pointing out the main features of the new dimensional FEA was performed simulating the doubler
geometries regarding the maximum Von Mises stresses plate discontinuity and the presence of the brackets
induced in the structure, the number of existing hot (Figure 12) or the bulb flat. Accordingly with Figure 12
spots, the maximum deflection value obtained and the and with the data presented in Table 6, both proposed
mass of the new bilge keel’s internal geometries solutions could be considered to reduce the stresses
developed. induced in the doubler plate connection to the ship’s
hull, ensuring the water tightness of the hull.
Table 5. Results obtained from the FEA of the new
internal bilge keel geometries.
Maximum Von Mises
Mass Nr.of hot
Deflection Stress
(kg) spots
(mm) (MPa)
Current Geometry 41,34 5 3.8 364 MPa

Geometry 1 39,44 3 249 MPa

Geometry 2 40,90 3 255 MPa

Geometry 3 37,77 6 246 MPa

≈3.3
Geometry 4 39,27 6 260 MPa

Geometry 5 37,84 3 230 MPa Figure 12. Two-dimensional FEA of the bilge keel
connection to the brackets
Geometry 6 39,33 3 248 MPa
Table 6. Stresses present in the bilge keel upper plating,
The analysis showed a significant reduction in the stress with the implementation of brackets and bulb flat.
level (a minimum reduction of 28% and a maximum
reduction of 37%). Of all the analysed geometries, Stress in Stress in Stress in Stress in
geometry number 5 presents the best alternative to the Point 1 Point 2 Point 3 Point 4
[MPa] [MPa] [MPa] [MPa]
existing geometry, because it’s definitely the geometry
that showed the lowest level of maximum induced
Current model 157 149 29 50,6
stress, 230MPa, together with the smallest number of
hot spots and a lower mass. Notwithstanding, Application of brackets 155 154 4,8 3,9
alternative geometries which include longitudinal Application of bulb plate 122 134 29,7 27,3
reinforcements, namely geometries number 6, 2 and 4
(Table 4), in this order, should be considered in case
bilge keel longitudinal strength needs to be enhanced. 6. FATIGUE ANALYSIS
These solutions also revealed a lower maximum
induced stress than the current geometry (Table 5).  A fatigue strength assessment of the redesigned doubler
plate connection to the ship’s hull (Figure 13) has been
5.2. Improved Design to Reduce the Stress in the Bilge made assuming the inexistence of the doubler plate and
Keel Connection to the Hull in accordance with the fatigue design curves included in
Germanischer Lloyd Rules [4], which state that for
Having been registered several cracks in the bilge keel structures submitted to cyclic loads resulting from
connection to the hull, two alternative solutions have waves or engines [4] and for infinite life, eq.2 must be
been designed, so as to reduce the maximum induced fulfilled.
stress value to the connections under study (Figure 11),
in order to eliminate the structural failures found.
Δσ Rc ⋅ f n ≥ Δσ ⋅ fa ⋅ Fdyn ( (2))
where ΔσRc is the corrected fatigue strength reference
value, fn is the correction factor considering the
allowable number of load cycles and the type of
spectrum applied and Δσ(fa.Fdyn) is the stress range due
to the dynamic loads applied in the assessed structure.

a) b)
Figure 11. Cross section view of the bilge keel
connection to the hull with
a) brackets. b) bulb flat
The proposed solutions consisted in the application of
Figure 13. Simplification of bilge keel to the hull
brackets inside the hull (Figure 11a), or a bulb flat along
connection with internal stiffeners (bulb flat).
The S-N curves (Figure 14) for the welded detail stress level induced by the presence of the bilge keel
represented in Figure 13, relative to two standard transversal reinforcement only.
classes of Spectra A and B [4], were corrected through
the application of a factor, fn, [4], where Spectrum A A 2D FEA performed to the doubler plate and the bilge
represents the denominated straight-line spectrum, keel, and between the doubler plate and the hull allowed
which is a typical stress range spectrum of seaway- to determine the maximum Von Mises stress present in
induced stress ranges [4] and Spectrum B represents a the welded joint due to the stress concentration
parabolic spectrum, which represents a Normal function introduced by the welded detail, fact which is in
density of probability type of stress range. The fatigue accordance with the failure observed. The application of
curves presented in Figure 14 are representative of the brackets or a bulb flat allowed to reduce the maximum
fatigue resistance of medium/high strength structural stress induced to 155MPa between the doubler plate and
steels used in shipbuilding [4]. the bilge keel.

Once known the maximum stress value present at the In order to propose a new solution capable of
cruciform joint (155 MPa, Table 6) it was assumed a preventing the structural failures identified in the
local stress of 170 MPa (Figure 14) due to the weld toe internal bilge keel structure, six alternatives geometries
stress concentration factor (Figure 13) for the stress to the internal structure of bilge keel were designed and
variation present in structure due to dynamic loads, Δσ analysed, maintaining unchanged the hydrodynamic
(fa.Fdyn). The fatigue life prediction can be observed in characteristics of the bilge keel. The best solution found
Figure 14. For the spectra B or A, which reflect the (geometry nr.5, Table 4) contains 3 hot spots regions
service conditions of the ship, the estimated lifetime is and a maximum Von Mises stress of 230MPa,
approximately 16 years and infinite lifetime, corresponding to a reduction of 37% in the stress
respectively. The cycle conversion was carried out once induced levels when compared to the current geometry.
known that 5x107 fatigue life cycles in a class A Notwithstanding alternative geometries which include
spectrum, is equivalent to a lifetime of 25 years with longitudinal reinforcements, namely geometries number
230 days per year of service at sea [4]. 6, 2 and 4 (Table 4), in this order, should be considered
in case bilge keel longitudinal strength needs to be
enhanced. These solutions also revealed a lower
maximum induced stress than the current geometry
(Table 5).

Finally, a fatigue life assessment analysis to the bilge

keel welded connection to the hull resulted in an infinite
lifetime for a Spectrum A type. Adopting a more
conservative approach, the fatigue lifetime for a class B
spectrum estimated a fatigue lifetime of 16 years.

REFERENCES

Figure 14. Fatigue life prediction for spectra A and B [1] ON THE STRUCTURAL DESIGN OF BILGE
KEELS. Mateus, António F. Hamburg, Germany :
2006. 25th International Conference on Offshore
7. CONCLUSIONS Mechanics and Arctic Enginneering.

This paper refers to an analysis performed to the [2] Det Norske Veritas, Offshore Standard DNV-OS-
structural failures found in bilge keels, with the B101, Metallic Materials, April 2009.
objective of presenting structural solutions to solve
them. [3] Grant, David James. Full Scale Investigation of
Bilge Keel Effectiveness at Forward Speed. April
It was confirmed, through the analyses carried out by 28, 2008.
FEM that the location of the internal failures in the bilge
keels coincides with high stress concentration regions. [4] Lloyd, Germanischer. Rules for Classification and
It was verified the existence of stress levels close to the Construction, III Naval Ship Tecnology. :
material yield strength and, consequently, some areas Germanischer Lloyd, 2004.
were identified where local plasticity can occur.
[5] Zhang, Guihua. Development of Numerical
The alignment between the bilge keel’s internal Approaches to Predict Ductile and Cleavage
structure and the hull transversal reinforcements results Fracture of Structural Materials. University of
in an increased and more concentrated stress level along Akron, December 2007.
the hull-bilge keel connection, when compared to the