Professional Documents
Culture Documents
Marine Structures
journal homepage: www.elsevier.com/locate/
marstruc
a b s t r a c t
1. Introduction
The stiffened plate, a major structural element of ships and offshore structures, is prone to damage
due to collision with other floating or fixed structures and falling objects. Rational design of stiffened
plates against such kinds of mass impacts must incorporate predictions of the probability of the
* Corresponding author. Tel.: þ82 52 259 2163; fax: þ82 52 259 2836.
E-mail address: srcho@mail.ulsan.ac.kr (S.-R. Cho).
0951-8339/$ – see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.marstruc.2008.06.003
S.-R. Cho, H.-S. Lee / Marine Structures 22 (2009) 84–95 85
occurrence of accidents, the extents of damage and the residual strength of the damaged structure [3].
Many researchers have performed experimental and theoretical studies on the behaviour of stiffened
and unstiffened plates subjected to lateral impulsive loadings. Nurick and Martin [9] and Nurick and
Shave [10] reported on the deformation and tearing of thin plates subjected to impulsive pressure
loads. Wierzbicki and Nurick [15] investigated the response of thin clamped plates to a localized
impulsive load imparted over a central region experimentally and theoretically to determine the
location of tearing failure and the critical impulse to failure. Lee and Wierzbicki [6,7] performed
a comprehensive analysis of deformation and fracture of thin plates subjected to localized impulsive
loading. They investigated many aspects of the problem including calibration for plasticity and fracture.
Park and Cho [11] proposed simple design formulae to predict the extents of damage of unstiffened and
stiffened plates under explosion loadings.
Samuelides [12], Zhu and Faulkner [16], and Cho et al. [1] reported results from experimental and
theoretical studies on the structural behaviour of unstiffened plates subjected to lateral collisions.
Wang [13] reviewed equations proposed to predict the plastic deformation of plates subjected to static
punching loads. Lee et al. [8] investigated the response of thin clamped plates subjected to static punch
indentation experimentally, analytically and numerically to determine the onset of fracture. Kaminski
et al. [5] summarized research on ship structures under lateral collisions. Wang et al. [14] reviewed the
publications on ship’s side collision with wedge-type or bulbous bow.
Even though the stiffened plate is a basic structural element of marine structures, results from
stiffened plate collision tests have not been reported in the open literature. Because ships and offshore
structures consist of stiffened plates, their overall response to lateral collision loading cannot be
considered without first studying the behaviour of stiffened plates subjected to lateral collisions.
Commercial numerical packages, such as DYTRAN or ABAQUS, are appropriate for the analysis of
stiffened plates subjected to collision, but their operation may be considered overly expensive and time
consuming for design purpose. Therefore, it is essential that any simple, design oriented methods for
stiffened plate collision analysis be developed.
The predictive capability of any commercial or in-house computer program or analytical method
should be substantiated with relevant test data before performing nonlinear dynamic analyses.
Reported in this paper are the results of lateral collision tests run on 33 stiffened plate models. Of the
plate models, three were cracked and one was torn. A simplified analytical method is proposed for the
prediction of the extents of damage to stiffened plates subjected to lateral collisions. In this method it is
assumed that all kinetic energy of the striker is dissipated as plastic deformation of the struck stiffened
plate. The plastic deformation consists of hinge rotation occurring at plate and stiffener flanges, the
membrane tension of plates and stiffener flanges and the shear deformation of stiffener webs. The
proposed method is substantiated with the test results. The mean ratios of predicted to actual lateral
damage for 29 models without crack or tearing damage was 1.138 with a COV of 19.9%. Using the test
data and the proposed analysis method, a criterion was developed to judge the occurrence of cracks.
Two kinds of collision testing machines were used in this study. Strikers lighter than 60 kg were
accelerated by springs and were run on a frictionless roller conveyer [1]. Strikers heavier than 250 kg
were accelerated using a pendulum type machine. The pendulum type collision testing machine is
composed of a pendulum, a pendulum hanging frame and a pendulum height controller (see Fig. 1).
The model support structure is made of 150 150 10/8 H-beam, which has relatively much larger
stiffness comparing with those of the models.
The mass of the striking pendulum can be increased up to 600 kg and the maximum collision speed
is about 3.5 m/s. The colliding velocity of the pendulum was measured using a timer, which starts and
stops using two different light sensors. The pendulum has a knife edge striker on its front face with
a height of 165 mm. The cross-section plan of the striker header made of 75 75 10 angle stiffened by
a flat bar of 10 mm thickness is provided in Fig. 2. The striker was released from a desired height to
produce an expected velocity.
86 S.-R. Cho, H.-S. Lee / Marine Structures 22 (2009) 84–95
Fig. 1. Overall view of pendulum type collision testing machine with a model in position.
Collision test models were fabricated using CO2 welding. The boundaries of all models were stiff-
ened by angles of 75 mm 75 mm 6 mm. In the tests models were firmly bolted to the support
structures. The material properties, elastic modulus and yield stress, were measured in tensile tests.
Five tensile coupons were cut from one parent plate sheet. The yield stress was determined to be equal
to 0.2% of the proof stress. The geometric and material properties of models are summarized in Table 1.
One or three stiffeners were welded onto each plate and the stiffener spacings were equal for all
models.
In order to obtain strain history information for struck models, strain gauges were glued onto each
of the six models, CS1, 5, 6, 8, 11 and 13. For each model, eight strain gauges were bonded to the back
face, where stiffeners were welded. Four gauges were bonded to the plate, three were bonded to the
stiffener flange and one was bonded at the mid-height of the stiffener web. Fig. 3 shows the locations of
strain gauges and a photograph of a strain-gauged model, CS6, is provided in Fig. 4. Strains were
recorded using amplifiers and an A/D converter. The data were acquired 10,000 times per second.
Table 1
Measured geometric and material properties of test models and lateral collision test results
In Fig. 5, the measured depth of the dent of model CS13 is visualized and a longitudinal cross-section
view is presented in Fig. 6. As can be seen in the figures, the depth of the dent increases linearly from
the boundaries, but then sharply increases in the middle part, which is where the striker hit. The
figures can provide useful information for any theoretical study. Photographs of damaged models CS1
and 8 are presented in Figs. 7 and 8. Model CS1, which is stiffened by inverted angles and is stiffer
relative to CS8, shows local deformation. However, model CS8, which is stiffened by flat bars and is
more flexible than model CS1, shows tripping of the middle stiffener that extends to the model corners,
quite far from the contact and yield lines.
Fig. 5. Plots of measured depth of dent for model CS13 (depth is amplified in the figure).
collided with the models off-center, the strain histories fail to show symmetric responses. In Fig. 9 the
records of two strain gauges are cut due to their excess preset limit.
In this paper, a simple analytical method for prediction of the extent of damage to stiffened plates
subjected to lateral collision loads is proposed. The assumptions adopted in the proposed method are
as follows.
- The kinetic energy of the striker can be dissipated by the plastic deformation of the struck stiffened
plate.
- The deformed shape of the struck stiffened plate can be represented in a manner consistent with
Fig. 10.
In hoping to make the derived equations more versatile, the deformed shape assumed in Fig. 10 is for
the collision by a rectangle striker header. If the header is a knife edge, then ac in the figure should be
null. In the derivation of the equation, it is also assumed that the collision is central, which may produce
most severe damage than others. Therefore, this assumption may be conservative for other cases.
The plate can absorb some portion of the kinetic energy of the striker through the rotation of plastic
hinges formed at yield lines and through the membrane plastic tension. The expressions for the energy
Fig. 6. Longitudinal cross-section view of measured depth of dent plots for model CS13 (depth is amplified in the figure).
90 S.-R. Cho, H.-S. Lee / Marine Structures 22 (2009) 84–95
Fig. 7. Photographs of damaged model CS1 (a) front view, (b) rear view, and (c) close-up of damaged part.
Fig. 8. Photographs of damaged model CS8 (a) front view, (b) rear view, and (c) close-up of damaged part.
S.-R. Cho, H.-S. Lee / Marine Structures 22 (2009) 84–95 91
Fig. 9. Strain history (a) model CS11 and (b) model CS13.
absorbed by the membrane tension and hinge rotation are given as Eqs. (1) and (2), respectively [1].
Dynamic tensile tests were not performed in this study. However, dynamic tensile tests have been
conducted on a material similar to those used for these tests. Their results, which were reported
elsewhere [1], confirmed that the Cowper and Symonds strain rate sensitivity equation can be
employed for the investigated steel. Therefore, in the derivation of the equations the dynamic yield
stress is approximated using the Cowper and Symonds [2] relationship.
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Utp ¼ 2ðUtI þ UtII Þ ¼ 2t sYDI ðac þ aÞ b2 þ d2 b þ sYDII ðbc þ bÞ a2 þ d 2 a (1)
Ubp ¼ 2 mpI ða þ ac Þq1 þ mpII ðb þ bc Þq2 þ 2mpIII ld ðq3 þ q4 Þ (2)
where a and b are width and height of the plate, respectively; ac and bc are width and height of the
striker header, respectively; t is plate thickness; Utp is dissipated energy by the membrane plastic
tension of the plate; Ubp is dissipated energy by the rotation of plastic hinges of the plate; d is maximum
permanent deflection of the plate; sYDI ¼ sp f1 þ ð3_ I =40:4Þ0:2 g, sYDII ¼ sp f1 þ ð3_ II =40:4Þ0:2 g,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
sYDIII ¼ 0:5ðsYDI þ sYDII Þ; sp is static yield stress of the plate; 3_ I ¼ 0:5Vi ð b2 þ d2 bÞ=ðbdÞ,
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
3_ II ¼ 0:5Vi ð a2 þ d2 aÞ=ðadÞ, mpI ¼ 0:25sYDI tp2 , mpII ¼ 0:25sYDII tp2 , mpIII ¼ 0:25sYDIII tp2 .
It is further assumed that the stiffeners can absorb energy through the rotation of plastic hinges
formed at yield lines, through the membrane plastic tension in the stiffener flanges and through the
plastic shear deformation of stiffener webs. The energy dissipated by the membrane plastic tension and
rotation of the plastic hinges in the stiffener flanges can be expressed as Eqs. (3) and (4), respectively.
ns
X qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
Utf ¼ 2 sYDs tf wf a2 d2i a (3)
i¼1
1
Ek ¼ MV 2 (5)
2 i
pffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffi pffiffiffiffiffiffiffiffiffi
Ep ¼ sYp t 2 ab þ sYs ns tw
2
ahw þ tf2 awf (6)
where Ek is the initial kinetic energy of the striker; Ep is the nominal energy absorbing capacity of the
stiffened plate.
It seems necessary to explain why Eq. (6) was chosen for the expression of the nominal energy
absorbing capacity of the stiffened plate. In the very beginning stage of collision where the elastic
behaviour is dominant, the bending deformation can contribute to the total strain energy of the struck
structure somehow. However, this portion becomes smaller and the membrane tension becomes
dominant when the lateral defection becomes larger. Therefore it was decided to express the plastic
capacity of the struck structure with a parameter which can represent the plastic membrane defor-
mation of the structure. Other types of equations were also tried, but Eq. (6) is quite simple yet showed
best correlation. Then, the ratio Ek/Ep can nominally represent the severity of the collision.
The stiffener web can absorb the energy through plastic shear deformation. Adopting the von Mises
yield criterion, the energy dissipated through the plastic deformation of the stiffener webs can be
written as Eq. (7).
ns
X
Usw ¼ 2 sdi tw hw (7)
i¼1
where tw is the thickness of the stiffener web; hw is the height of the stiffener web; di is the maximum
pffiffiffi qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
permanent deflection of the ith stiffener; s ¼ 1= 3ð s2YDs s2 Þ, s ¼ sYDs ef0:1ðEk =Ep Þg .
The total energy dissipated by the stiffened plate can be expressed as Eq. (8).
S.-R. Cho, H.-S. Lee / Marine Structures 22 (2009) 84–95 93
When the total dissipated energy equals the kinetic energy of the striker (Ek), then the plastic defor-
mation stops and the lateral deflection at that moment is the predicted depth of dent.
4. Discussion
In collision analysis, it is very important to judge the initiation of crack damage in the struck
structure. The non-dimensional dent depth (d/t) of test results is depicted in Fig. 11 against the energy
ratio (Ek/Ep). As can be seen in the figure, the models having through-thickness crack or tearing damage
provide the upper bound of the data trend line.
A lower bound equation can be derived from the figure and represented as Eq. (9). The lower bound,
however, seems to be somewhat pessimistic, especially in the high energy region. It is sensible to use
this lower bound equation until further test data with through-thickness crack or tearing damage are
available.
qffiffiffiffiffiffiffiffiffiffiffiffiffi
d=t ¼ 8 Ek =Ep (9)
Of course, sophisticated analyses seem to be necessary to predict the initiation of crack damage
considering dimensions and material properties of striking and stuck structures, locations of struck
zones and so on. In this study, however, a simple empirical expression obtained by regression analysis
of test data is tried to be derived with which a simple check of the occurrence of crack can be made.
In Table 2, predicted dent depths for 33 models are presented together with their respective energy
ratio (Ek/Ep) and the ratio of predicted to actual dent depth (Xm). The mean of Xm for 33 models is 1.085
with a COV of 23.7%. When excluding the four models (CS1, 5, 6 and 8) with through-thickness crack or
tearing damage, the mean is 1.138 with a COV of 19.9%. As compared to the accuracy of good prediction
methods for static structural problems, the aforementioned COVs are a bit larger. However, considering
the high degree of uncertainty associated with structural impact problems, the proposed method may
be interpreted to be reasonably accurate given its simplicity. If the predicted damage value is larger
than the value provided by Eq. (9) after performing damage prediction analysis using the proposed
25
20
15
uncracked
10
partially cracked
completely cracked
torn
5 eqn(9)
0
0 1 2 3 4 5 6 7
Fig. 11. Plots of non-dimensional dent depth of test results against energy ratio (Ek/Ep).
94 S.-R. Cho, H.-S. Lee / Marine Structures 22 (2009) 84–95
Table 2
Comparison of damage predicted using the analytical method with damage resultant from testing
method, the possibility of the occurrence of crack damage should be considered. This paper provides
not only a method for predicting the extent of damage of stiffened plates subjected to a lateral collision,
but also a criterion to estimate the occurrence of crack damage.
5. Conclusions
In this paper, results of lateral collision tests on 33 stiffened models are reported. Among the
models, two were partially cracked, three had through-thickness cracking and one was torn. These
results can help to substantiate any commercial analysis package or newly developed analysis method.
The proposed damage prediction method provides a mean of 1.085 for Xm of 33 models together with
a COV of 23.7%. For the 29 uncracked models, the mean is 1.138 with a COV of 19.9%. Further
improvement may be achieved by considering contact location in the analysis.
The assumed shape of the deformed plate in this study is composed of flat segments. In the actual
deformed models, however, there is a local dent where the striker impinged (see Fig. 6). It seems likely
that if the local dent is considered in assuming the deformation shape, improved predictions can be
expected. Of course, if more realistic deformed shape is assumed the equations will be more
complicated.
S.-R. Cho, H.-S. Lee / Marine Structures 22 (2009) 84–95 95
Further collision tests are necessary especially where higher kinetic energy of the striker is
employed to gain an expectation of crack or tearing damage. The crack occurrence criterion, Eq. (9),
may improve with an increase in available test data with crack or tearing damage.
The equations proposed in this paper are substantiated with or derived based upon the test results
on small scale model. However, when subjected to impact loads, small scale models are stronger than
full-scale prototype [4]. It is also necessary to perform further research to quantify the scale effect in
structural impact problems.
Acknowledgements
The work reported in this paper was supported by 2003 research fund of the University of Ulsan.
References
[1] Cho S-R, Kim I-W, Lee S-B. Experimental investigations on the plastic damage of plates due to lateral collisions. Society of
Naval Architects in Korea (SNAK) 2002;6(3):1–12.
[2] Cowper GR, Symonds PS. Strain-hardening and strain-rate effects in the impact loading of cantilever beams. Tech. Rep. No.
28, Brown University; 1957.
[3] Frieze PA, Cho S-R. Dynamic impact to tubulars and their residual strength. In: Proceedings of the fourth international
symposium on practical design of ship and mobile units (PRADS ’89), Varna, 1989, p. 50/1–50/7.
[4] Norman J. Structural impact. Cambridge: Cambridge University Press; 1989 [Chapter 11].
[5] Kaminski ML, Amdahl J, Fasano E, Frieze PA, Gordo JM, Grundy P, et al. Report of Technical Committee III. 1: ultimate
strength. In: Proceedings of the 14th ship and offshore structure congress. Elsevier; 2000. vol. 1, p. 291–99.
[6] Lee Y-W, Wierzbizki T. Fracture prediction of thin plates under localized impulsive loading. Part I: dishing. International
Journal of Impact Engineering 2005a;31:1253–76.
[7] Lee Y-W, Wierzbizki T. Fracture prediction of thin plates under localized impulsive loading. Part II: discing and petalling.
International Journal of Impact Engineering 2005b;31:1277–308.
[8] Lee Y-W, Woertz JC, Wierzbizki T. Fracture prediction of thin plates under hemi-spherical punch with calibration and
experimental verification. International Journal of Impact Engineering 2004;46:751–81.
[9] Nurick GN, Martin JB. Deformation of thin plates subjected to impulsive loading – a review: part I, theoretical consider-
ations/part II, experimental studies. International Journal of Impact Engineering 1989;8(2):159–86.
[10] Nurick GN, Shave GC. The deformation and tearing of thin square plates subjected to impulsive loads – an experimental
study. International Journal of Impact Engineering 1996;18:99–116.
[11] Park B-W, Cho S-R. Simple design formulae for predicting the residual damage of unstiffened and stiffened plates under
explosion loadings. International Journal of Impact Engineering 2006;32:1721–36.
[12] Samuelides E. Structural dynamic and rigid body response coupling in ship collision. PhD thesis, Glasgow University; 1984.
[13] Wang G. Some recent studies on plastic behavior of plates subjected to large impact loads. Journal of Offshore Mechanics
and Arctic Engineering 2002;124:125–31.
[14] Wang G, Spencer J, Chen Y. Assessment of a ship’s performance in accidents. Marine Structures 2002;15:119–38.
[15] Wierzbizki T, Nurick GN. Large deformation of thin plates under localized impulsive loading. International Journal of
Impact Engineering 1996;18(7–8):899–918.
[16] Zhu L, Faulkner D. Dynamic inelastic behaviour of plates in minor ship collisions. International Journal of Impact
Engineering 1993;15(2):165–78.