CRACK PROPAGATION IN NAVAL ALUMINUM PANELS
K.P. Galanis and V.J. Papazoglou
Shipbuilding Technology Laboratory
School of Naval Architecture and Marine Engineering
National Technical University of Athens
9, Heroon Polytechniou Ave.
GR-157 73 Zografou Athens
GREECE
galanis@naval.ntua.gr, papazog@deslab.ntua.gr
ABSTRACT
The shipbuilding industry is seeking to achieve advanced and more efficient concepts and designs for naval vessels with
improved safety and performance using optimized structural design. To increase the vessel’s survivability, in-depth
understanding and vast experience in studying the methodology and dynamic effects of complex damage mechanisms on
marine structures and ship systems is required. Damage prediction models should be created to enhance the ability of the ship
to withstand all types of loads. Verification and validation of modeling and simulation is required for the models to be
functional. The modeling could be either numerical or physical. All these will eventually lead to the development of design
procedures and criteria, which can create affordable designs. Structural failure due to extreme loads is most often caused by
fracture. It is a common practice to design ships with adequate resistance to yielding, buckling and fatigue, but not fracture.
Consequently, adequate methods and procedures to design ships against fracture have not been developed. Even more
importantly, a fundamental understanding of mechanisms and mechanics of fracture is lacking. Current trends for lighter
structures increased the use of aluminum for marine applications. The present paper is focused on the mapping of the crack
path in naval aluminum panels by examining different types of specimen dimensions, type of stiffeners, stiffener configuration,
notch geometry (initial length and tip) and displacement rate. The experiments were performed on naval aluminum Compact
Tension (CT) specimen under quasi-static loading.
Introduction
Ships represent some of the most complex thin-shelled structures seen in engineering, yet until recently, ship structural
analysis was based upon considerable empiricism. The increased use of the finite element method, together with high-speed
computers, has allowed more rational approaches to ship structural analysis and design to be adopted. A ship structure can be
regarded as a complex assemblage of continuous stiffened plates. One of the most common modes of failure in such plates is
fracture. Crack propagation and, especially, arrest in naval aluminum panels, however, has not been an area of in-depth
research. As a consequence, the increased use of aluminum in the shipbuilding industry is highly dependent on the
understanding of the fracture mechanisms that govern naval aluminum panels, given that the vessels operate in extreme
environmental and functional conditions that may lead to the loss of the entire structure.
The crack arrest phenomena in cylindrical containers and pipes have been extensively studied in relation to the pressure
vessel technology, e.g. Freund et al [1] and Zhuang and O’Donoghue [2]. Similar analyses, though, are not readily available
for the case of the tearing fracture mode in plates, in which case the mechanics of the crack arrest process must be well
understood. For example, the webs or girders in the Advanced Double Hull (ADH) structure seem to offer a formidable
obstacle for a propagating crack in the transverse direction. Over the last decade the US Navy has been actively pursuing
research of the ADH for naval combatants [3].
The ADH structure is comprised of inner and outer shells connected longitudinally with plate girders. This arrangement results
in a string of long boxes or cells (Figure 1). ADH structures were successfully optimized for normal operational loads (still
water and wave loads) and Design Guidelines for the Unidirectional Double Hull Structure were developed [4].
Figure 1. Stainless steel ADH mid-body section
Tearing fracture in stiffened plates was considered up to now to be arrested for a while at the foot of the stiffener, and then be
re-initiated on the other side of the stiffener; eventually the longitudinal stiffener would fracture as well. A test on a quarter
model of splitting damage of a longitudinally stiffened, double hull, which was performed by Rodd [5], gives only a glimpse on
this important phenomenon. It is still unknown, however, what aspect of the design (strength of the fillet weld, relative
thickness of the bare plate and stiffener, height of the stiffener or hull separation) will stop the crack, re-direct it or overcome
the obstacle (Figure 2). Recently, Paik and Thayamballi [6] have performed some preliminary work on the estimation of the
strength of cracked aluminum unstiffened plates.
Figure 2. A conceptual sketch of the crack arrestor. Stopping the crack (left), redirecting (center), and passing through (right)
In the “conventional” ADH design, there is no mechanism to stop tearing fracture in the longitudinal direction. The concept of
welded (or even hot rolled) buffer strips should thus be investigated, as illustrated in Figure 3. Finally, the process of plate
tearing in the diagonal direction with respect to the longitudinal stiffener must be fully understood. It is expected that cracks will
run in a “zig-zag” pattern. For the purpose of designing optimum blast resistant structures, the ultimate intellectual challenge
will be to develop a general concept of crack resistant core of sandwich structures. It is unknown at this point which design will
offer better resistance to tearing; a uniform plate or a slightly thinner plate with integrated ribs of the same weight.
Figure 3. An example of a hot-rolled ribbed plate with integrated crack arrestors
As a first step and in order to map the crack path in naval panels, a series of experimental tests and numerical analyses were
performed based on the standard Compact Tension (CT) specimen. For this case, the aluminum alloy grade 1561 (relevant to
the ISO 209-1 based AlMgMnZr alloy) of 4 mm thickness was selected which has a density of 2.65*103 kg/m3 and yield stress
of 190 MPa. Aluminum-magnesium alloy 1561 is among the average strength heat non-hardenable alloys. Being heated by
welding, alloys of this group do not reduce their strength properties in the heat-affected zone. Several uniaxial tests of
“dogbone” specimens were initially conducted to verify the material properties and to examine the failure mode (slant fracture)
of this naval aluminum alloy (Figure 4). These results will also be used for calibrating the numerical simulations, as part of
future research efforts. The CT specimens were then tested by varying several parameters, such as the structural
configuration, the geometries of the specimen and crack tip, the displacement rate, the stiffener type and configuration
(extruded and welded) and the crack length.
Figure 4. Mode of fracture (slant fracture)
Unstiffened Compact Tension Specimens
Introduction of a notch into the gage length of a metallic tensile test specimen increases the tensile strength above that
measured on an identical metallic test specimen without a notch. The increased tensile strength of the notched specimen is
caused by a subtle distortion of the applied uniaxial stress, resulting in a localized triaxial stress state at the root of the notch.
Evaluation of research materials may be performed using standard or nonstandard test specimens, depending on the amount,
size and shape of material available. Quasi-static tests were conducted on notched CT specimens of 4 mm thickness (with no
precracking). The specimens had the dimensions presented in Figure 5. The tests were conducted at room temperature and at
atmospheric pressure, using the installed MTS servo hydraulic universal testing machine of the Shipbuilding Technology
Laboratory, National Technical University of Athens. During testing the load-displacement relationship was measured through
a computerized data acquisition system, whereas at the same time the crack propagation pattern was traced through
consecutive photographs taken at predetermined short intervals.
Figure 5. CT specimen dimensions in mm
Thirteen unstiffened CT specimens were tested and in all cases the crack propagated perpendicular to the loading direction. It
was observed that the crack initiated at a 45 degree angle at the notch and that the parameters examined (notch length and tip
geometry, displacement rate) did not affect the crack direction (Figure 6).
Figure 6. Unstiffened CT specimens tested under varied displacement rate and crack length and tip geometry
Stiffened Compact Tension Specimens
A ship structure is composed almost entirely of orthogonally (longitudinally and transversely) stiffened plating. Both the plating
and stiffeners must be designed to sustain the working loads. Plate panels in the ship are welded around their periphery to
stiffeners or adjacent panels. Fatigue tests are addressing the properties of local structural issues, such as capacity of plating
between stiffeners, the proportions of the flange and web elements of stiffeners, and the fatigue strength of a welded detail.
Local structural issues like these should, however, be understood before the larger, general structure can be designed. As the
overall structure increases in size, the size, shape, and length of the members may be limited in order to control buckling of
the main load carrying members (Figure 7). Therefore, grillage test structures, containing multiple longitudinal and transverse
stiffeners, have to be evaluated for catastrophic modes of buckling failure and to define margins of safety for design against
ultimate failure.
Figure 7. Various types of beam members (stiffeners)
The most common techniques used in the aluminum shipbuilding industry for naval panel construction is either welded
stiffeners or extruded ones (Figure 8). In the tests performed (again on CT specimens of the same dimensions as before, but
now having a welded or extruded stiffener), it was observed that the crack did not propagate through the stiffener at the T-type
extruded specimens (shown on the right of Figure 7 and produced during the manufacturing process, meaning physically
attached to the base plate). Additionally, and based on the base plate material behind the stiffener web, a “zig-zag”
phenomenon was observed at around 1/3 of the distance between the initial notch and the stiffener web. While the crack
reached the extruded stiffener’s web, a delay was noticed, meaning that the crack “tried” to propagate through the stiffener,
but did not succeed. On the other hand, in two of the four cases of the I-welded specimens (shown on the left of Figure 7 and
butt welded on the base plate following a standard welding procedure), the crack managed to propagate through the web and
behind the stiffener. The sequence of the crack propagation can be categorized in the following three steps: (a) through the
weld, (b) on the web of the stiffener, and (c) at the rear plate material. Finally, the displacement rate and the initial notch
geometry (crack length and different crack tip) did not seem to significantly affect the crack path.
Figure 8. Stiffened CT specimens tested, I-type welded stiffener (left) or T-type extruded stiffener (right)
Results and Analysis
Structural damage can be reduced by the use of new materials, novel structural configurations, different design, manufacturing
methods and improved computational techniques. It is clear that this cannot be accomplished by a single improvement but
rather with a combination of various measures. The experiments conducted revealed a strong relationship between crack path
and the presence of stiffeners. Table 1 summarizes the main dimensions of the stiffened CT specimens, as well as the crack
lengths observed. The thickness of the plates was 4.1 mm. For the extruded T-type stiffeners the web height was 4.1 mm, the
web thickness was 3 mm, the flange length was 3 mm and the flange thickness was 4 mm. For the I-welded stiffeners the web
height was 4.1 mm and the web thickness was 3 mm.
Specimen
Description
Length
(mm)
Width
(mm)
Disp. Rate
(mm/min)
Crack
length
(mm)
Rear
Plate
(mm)
CT1E
CT extruded
167.0
120.0
0.5
55
65.0
CT2E
CT extruded
166.0
121.5
1.0
71
20.0
CT3E
CT extruded
167.0
120.0
1.5
71
20.0
CTEB
CT extruded
114.2
120.3
1.0
55
0.0
CT1W
CT welded
168.0
118.0
1.5
71
20.0
CT2W
CT welded
171.0
118.0
0.5
71
10.0
CT3W
CT welded
170.0
119.0
1.0
71
20.0
CT1WB
CT welded
111.9
117.8
1.0
55
0.0
CT2WB
CT welded
114.2
119.3
1.5
55
0.0
Table 1. Dimensions and measured crack lengths of the stiffened CT specimens
It was observed that the major parameter affecting the crack pattern is the plate material at the rear part of the stiffener, as
presented in Figures 9 and 10. Note that the specimens marked with the letter “B” at the end (meaning no rear plate material)
require the least load, whereas the crack approaches the foot of the stiffener’s web. Most of the specimens tested that meet
similar dimensional characteristics can be grouped together, since their load-displacement curves revealed low deviation
among them.
20
CT1W
18
CT2W
16
CT3W
14
Force [kN]
CT1WB
12
CT2WB
10
8
6
4
2
0
-1
-2
4
9
14
19
24
Displacement [mm]
Figure 9. Load vs. displacement curves for CT specimens with I-type welded stiffener
The crack managed to propagate through the web of the stiffener in two cases (CT2W and CT2WB) and failed to do so in all
other tests including both the extruded and welded CT stiffened specimens. The main phenomena observed for the extruded
CT specimens are the following: (a) the crack delayed its propagation the closer it was reaching the stiffener and accelerated
while it turned parallel to the stiffener, (b) the crack followed the “zig-zag” pattern at half the distance between the crack
initiation and the stiffener, and (c) the crack failed to propagate through the stiffener in all cases examined. Finally, the
displacement rate did not seem to affect significantly the crack path in the cases examined.
20
CT1E
18
16
CT2E
Force [kN]
14
CT3E
12
10
CTEB
8
6
4
2
0
-1
-2
4
9
14
19
24
Displacement [mm]
Figure 10. Load vs. displacement curves for CT specimens with T-type extruded stiffener
Conclusions
Naval surface ships and submarines must be designed to survive exposure to extreme loading conditions from impact and
explosions. Various structural failure modes contribute to the loss of integrity of naval vessels subject to blast loading; these
being dependent on material selection and structural configuration. Modeling the vessel’s response encompasses material
constitutive equations, fracture and damage mechanics, nonlinear dynamics simulation codes and structural finite element
analyses. The advancement and integration of these disciplines are necessary in order to provide a design framework for
developing optimum structural configurations and materials. Fracture is a multifaceted problem that spans lengths covering six
orders of magnitude. It starts at a micro scale length with the formation, growth and linkage of microvoids from which a unit
material volume is composed, and then continues to the thickness of the structure and the widths of a typical stiffening
element. Cracks initiated in this way propagate over the widths of a typical panel and even further across bulkheads and
decks, separating adjacent bays. In order to cover this broad range of failure mechanisms and length scales, a system
approach is necessary because structural failure must be dealt with all the way from the initiation site to the damaged state of
a hull girder.
The overall objective of this study has been to understand the mechanics of fracture that govern the structural response of
stiffened naval panels. The work has primarily been focused on high-speed crafts built with lightweight materials such as naval
aluminum.
It was observed that, when the crack propagates perpendicular to the loading direction, its direction is strongly dependent on
the presence of the stiffener, the stiffening configuration, type of stiffener and the material behind the stiffener. There was no
crack propagation through the stiffener in the case of extruded specimens. On the other hand, and under similar conditions,
the crack propagated through the stiffener when it was welded on the plate. The displacement rate and the crack tip geometry
seem not to significantly affect crack initiation and propagation. Four evolution phases were noticed: (a) crack initiation and
propagation up to the web of the stiffener, (b) crack propagation through the foundation of the stiffener, (c) crack propagation
on the stiffener, and (d) crack link-up and propagation at the plate behind the stiffener, with parallel propagation on the web of
the stiffener. The results of these experiments is expected to lead to the creation of a fracture test database on continuous
aluminum panels and the formulation of appropriate criteria for cracked aluminum panels in order to extend the service of
marine vehicles and structures. Future research efforts will be focused on repeating the experiments and on performing similar
ones on intermediate size specimens to examine the scaling effect. Finally, calibrated numerical simulation will provide
comparable results, which, we believe, will emphasize the effect of each parameter on the crack propagation and arrest
phenomena.
References
1.
2.
3.
4.
5.
6.
Freund, L.B., Li, V.C.F. and Parks, D.M., “Analysis of a Wire-wrapped Mechanical Crack Arrester for Pressurized
Pipelines”, Journal of Pressure Vessel Technology, Trans. of the ASME, 101(1), 51-58 (1979).
Zhuang, Z. and O’Donoghue, P. E., “The Recent Development of Analysis Methodology for Rapid Crack Propagation and
Arrest in Gas Pipelines”, Int. Journal of Fracture, 101(3), 269-290 (2000).
Beach, J., “Advanced Surface Ship Hull Technology – Cluster B”, Naval Engineers Journal, 103(6), 27-37 (1991).
Sikora, J., et al, “Advanced Double Hull Structural Design Technology”, Naval Engineering Journal, 109(5), 117-128
(1997).
Rodd, J.L., Phillips, M. P. and Anderson, E.D., “Stranding Experiments on Double Hull Tanker Structures”, In Proceedings
of the Advanced Double Hull (ADH) Technical Symposium, Paper No2, Gaithersburg, MD, (1994).
Paik K.P. and Thayamballi A.K., Ultimate Limit State Design of Steel-Plated Structures, John Wiley & Sons Ltd. (2003).