Sei sulla pagina 1di 11

Applied Ocean Research 26 (2004) 298308 www.elsevier.

com/locate/apor

Longitudinal strength of a high-speed ferry


T.E. Schellin a,*, A.Perez de Lucas b
a b

Germanischer Lloyd, Vorsetzen 35, 20459 Hamburg, Germany NAVANTIA (former IZAR), Velazquez 132, 28006 Madrid, Spain Available online 10 October 2005

Abstract The longitudinal strength of the high-speed ferry was investigated by subjecting the ships hull girder to long-term loads obtained from a frequency-domain panel code. Prior to the statistical analysis, linearly computed transfer functions were corrected for nonlinear effects, yielding two sets of transfer functions valid for different wave amplitudes. One set corresponded to the hogging condition; the other set, to the sagging condition. Two regular equivalent design waves were specied that resulted in loads representing the most severe global design load conditions. The still-water loading condition, yielding a still-water vertical bending moment in hogging, was superimposed on the wave-induced loads to obtain the total (design) loads in hogging. For the sagging condition only, additional impact-related loads were superimposed to obtain the total (design) loads in sagging. A nite element model of the ships structure was subjected to pressure distributions according to the two regular design waves. For comparison with classication society rule values, a simple beam theory strength analysis of the ships midship section was performed rst, and then another nite element analysis was carried out, whereby the imposed loads were tuned to the rule values of vertical bending moments. Rule-based magnitudes of nominal maximum longitudinal stress deviated signicantly (2539%) from comparable stresses obtained by the panel code based nite element analysis. However, stresses obtained from the rule-based nite element analysis agreed more favorably, especially in hogging. In the uppermost deck, for example, the panel code based compressive stress was only 9% larger than the comparable stress from the rule-based nite element analysis. q 2005 Elsevier Ltd. All rights reserved.

1. Introduction Under the European research project WAVELOADS (advanced methods to predict wave-induced loads for highspeed ships) practical tools were developed to compute wave-induced global loads on high-speed monohulls. Two alternative numerical methods were investigated. One method was based on a linear three-dimensional radiation/diffraction Green function formulation in the frequency domain that accounts for forward speed effects ([1]). The other method was based on a nonlinear time domain strip theory, taking into account the most dominant nonlinearities associated with the vertical responses (i.e. [12]). The research work also encompassed model tests of three fast monohulls in a seakeeping laboratory to obtain the necessary experimental data to validate the numerical methods and to test the limits of their application regarding the speed of advance. [2] presented a summary of the project results. Their results also include
* Corresponding author. Tel.: C49 40 36 149 312; fax: C49 40 36 149 7660. E-mail addresses: schn@gl-group.com (T.E. Schellin), aplucas@izar.es (A.P. de Lucas).

0141-1187/$ - see front matter q 2005 Elsevier Ltd. All rights reserved. doi:10.1016/j.apor.2005.08.004

the specication of global design loads based on rst principles as well as a structural strength analysis of a high-speed ferry, one of the ships investigated under the project. The strength analysis of this ferry is presented in this paper. The gross global load effects of waves are reected by the bending moments they cause on the ships hull. Thus, in this study the midship vertical bending moment was considered to be the most important global load parameter. A nite element model of the ships structure was constructed to perform a longitudinal strength analysis. Pressures resulting from the seakeeping analysis were transformed to nodal forces and then applied to the nite element model, thereby simulating the corresponding pressure distribution of two equivalent regular design waves for the hogging and sagging conditions. These design waves represented the most severe global hull structural response. From a practical standpoint, it was advantageous to use the frequency-domain technique to compute the wave-induced hydrodynamic pressures. The resulting pressures were integrated to obtain transfer functions of the vertical midship bending moments, and these bending moments were then statistically analyzed to obtain long-term (design) loads. The frequency domain technique used ([3,4]) was a panel method that relies on the zero-speed Green function, which meant that

T.E. Schellin, A.P. de Lucas / Applied Ocean Research 26 (2004) 298308

299

computations were performed according to the so-called encounter-frequency approach. The application of this linear method is restricted to small amplitude waves, resulting in equal magnitudes of vertical bending moment in the hogging and sagging conditions. Unrealistic wave-induced loads may result if major nonlinearities, such as the effects due to bow and stern are, are not accounted for. Therefore, the linearly computed, wave-induced hydrodynamic pressures were corrected for nonlinear effects according to the procedure developed by [5] and applied by, e.g., [6]. This procedure was rst proposed by [7] and further extended by [8]. The correction yielded two sets of transfer functions. One set corresponded to the hogging condition; the other set, to the sagging condition. The resulting transfer functions were valid only for specic wave amplitudes. Although nonlinear wave-induced global load effects were considered, a linear analysis was used to select the equivalent regular design waves. The justication for this approach was based on the study by [9]. They demonstrated that critical wave episodes that produce the largest linear responses can be expected to also produce the largest nonlinear responses. The numerically predicted wave-induced global loads were compared to results from classication society rule values, here calculated according to the requirements of [10] Rules for High-Speed Craft. 2. Ship particulars and wave climate The high-speed ferry represents current trends for newbuildings of fast passenger carrying monohulls in that it features high speed, small draft, light weight, and aluminum construction. Main particulars of the ship are listed in Table 1; a body plan is shown in Fig. 1. The ship was designed for restricted open-sea service, that is, for cruise mode operation in sea areas with a signicant wave height of not greater than 4.0 m for an average of not more than 10% of the year. According to classication society rules, this restriction denes the design vertical acceleration at the ships center of gravity and corresponds to an average of the 1% highest vertical accelerations in the most severe sea conditions expected. The ship was assumed operating the Western Mediterranean Sea. The corresponding wave data, adjusted in accordance with the above open sea service restriction, are summarized in
Table 1 Principal particulars of the high-speed ferry Length overall Length between perpendiculars Length at design waterline Breadth overall Draft Displacement Block coefcient Vertical design acceleration at center of gravity Vertical design acceleration at forward perpendicular Service speed Froude number 125.0 m 110.0 m 107.8 m 18.7 m 2.52 m 1796 t 0.6 0.94 g 1.87 g 39 kn 0.61

Fig. 1. Body plan of the high-speed ferry.

Table 2. These wave data are given as probabilities of occurrence of short-term sea states, here characterized by the signicant wave height and the zero up-crossing wave period ([11]). 3. The nite element model The longitudinal structural strength assessment employed the general purpose nite element code MAESTRO 7.1 that was part of the program system VERISTAR. For this purpose, a nite element mesh was constructed to idealize the highspeed ferrys structure. The nite element model, symmetrical about the ships center line, idealized the entire main load carrying structure of the hull. The model consisted of about 22,500 nodes and 15, 900 elements. Structural components included strake panels, frames, beams, columns, and plate elements. All stiffened plates and longitudinal girders were modeled using four-noded orthotropic shell quadrilateral elements, while two-noded beam elements represented the transverse beams. Pillars were modeled using two-noded eccentric beam elements. Bulb frames were modeled as L-beams of equivalent cross-sectional areas and properties. The complete nite element model is seen in Fig. 2. The structural model, consisting of two substructures, extended over the ships length. Substructure 1 (between frames 6 and 54) contained nine modules; substructure 2 (between frames 54 and 106), six modules. (The aft perpendicular was located at frame 0 and the forward perpendicular at frame 99.) Generally, modules were dened between two bulkheads. Three modules dened the structure between the aft perpendicular and frame 12. Four modules were located between frames 12 and 30 and included the vehicle ramps and the engine room. Seven modules dened the structure up to the forward perpendicular, and one module dened the structure beyond that location. Decks were dened extending over the ships length, and a deck bridge was added to close the upper side of the ship structure. Boundary conditions were specied to restrain the model from moving. This meant that some motions of certain nodes had to be inhibited. At the bottom, the center node located at

300

T.E. Schellin, A.P. de Lucas / Applied Ocean Research 26 (2004) 298308

Table 2 Western Mediterranean wave data for operation under OC 3 service restriction Signicant wave height (m) !1.0 1.02.0 2.03.0 3.04.0 Sum over all heights Zero-upcrossing period (s) 3.5 0.063 0.016 0.003 0.001 0.083 4.5 0.147 0.101 0.034 0.009 0.291 5.5 0.107 0.145 0.069 0.025 0.347 6.5 0.037 0.018 0.053 0.024 0.194 7.5 0.007 0.025 0.021 0.012 0.065 8.5 0.001 0.005 0.006 0.004 0.017 9.5 0.000 0.001 0.001 0.001 0.003 Sum over all periods 0.362 0.374 0.187 0.077 1.000

frame 0 was held xed; at the keel, the center node located at frame 79 was restricted from moving vertically; and at deck 3, the center node situated at frame 0 was restricted from moving horizontally. All other nodes were free to move in all planes. All masses and loads were balanced until reaction forces at model boundaries reached acceptable minima.

4. Still-water loads and wave-induced loads 4.1. Still-water loads With the aid of NAPA software, the ships static balanced vertical loading condition in still water was obtained. Static loads consisted of buoyancy forces and gravity forces, or weights, in the full-load condition. The longitudinal load distribution subjected the hull girder to a vertical still-water bending moment in hogging as shown in Fig. 3. At the ships midship section, located 42.3 m from the aft perpendicular, the resulting still-water bending moment was MSWZ40,935 kNm. 4.2. Wave-induced loads For the hydrodynamic load computations the ship was treated as a rigid body, i.e., exibility of the hull was not included. The wetted surface of the hull was subdivided into a nite number of small surface panels. Pressures at these panels approximated the pressure distribution on the hull. To extrapolate wave-induced pressures up to the wave contour, the input also had to include surface panels above the design waterline. Sufcient accuracy was achieved by discretizing the ships hull into 5650 at triangular and quadrilateral surface panels; 1636 of these panels were situated below the design

waterline. These surface panels were dened by 5815 nodes located at the panel vertices. The ships mass was idealized by 117 nodal masses located on the ships center plane. These nodal masses were specied according to the ships weight distribution. The discretization of the high-speed ferrys hull surface, including the ships the center plane, is shown in Fig. 4. To account for speed reductions in higher waves, computations were performed for the reduced ship speed of 30 knots, corresponding to a Froude number of 0.40. The wave induced pressure acting on a section of the ships hull comprised a hydrostatic part and a hydrodynamic part. The hydrostatic part was determined from the hydrostatic head up to the momentary still-water level of the ship, which was a function of the ships momentary position. The hydrodynamic part consisted of the linearly computed hydrodynamic pressures up to the still-water level and a nonlinear extension to account for pressures caused by wave loading in the region
Weight distribution Still-water bending moment

30 25

40000 35000 30000 25000 20000 15000

20 15 10

10000 5 0 10 0 5000 0 10 20 30 40 50 60 70 80 90 100 110 120

Length coordinate [m]

Fig. 3. Longitudinal weight distribution and still-water vertical bending moment.

Fig. 2. Finite element model of the ships structure.

Fig. 4. Surface discretization of the hull.

Still-water bending moment [kNm]

35

45000

Weight [t/m]

T.E. Schellin, A.P. de Lucas / Applied Ocean Research 26 (2004) 298308

301

a wave crest exceeds this value, the maximum wave elevation reaches its limit and the wave breaks. In this case the dynamic pressures is given by 1 ps Z pWL K rgs cos a 2 (3)

Solving this quadratic equation for z while setting sZz and psZ0, which are the boundary conditions for the wave contour, yields the wave elevation: r g cos a g cos a 2 pWL K K 2 zZ (4) 2 2u 2u2 ru
Fig. 5. Schematic of pressure distribution on ship cross-section.

between still-water level and wave contour. Fig. 5 shows a schematic diagram of the pressure distribution on a cross section of a ships hull positioned in a wave crest. Hachmanns method denes a time harmonic velocity potential with amplitude f and encounter frequency u to describe the two-dimensional ow in the vicinity of the stillwater level at the side of the ship. Applying Bernoullis equation to express pressure ps in the plane r,s tangent to the ships side yields  2  2  ps vf 1 vf vf C C C C gs cos a Z const: (1) vt 2 vr vs r where r is the density of water, g is the acceleration of gravity, and t is time. Coordinate r proceeds along a tangent at the ships still-water level, and coordinate s, orthogonal to r, extends from the still-water level positive upward along the side of the ship. The origin of the r,s coordinate system is located at the intersection of the still-water level and the considered ship section. Angle a, measured between the vertical and the ships side, designates the are of the ship section. The following relationship results for the pressure ps along the side of the ship: ps Z pWL Krg cos aKu2 zs (2)

The maximum surface wave elevation occurs at the crest when pWLZr(g cos a/2u)2 and, half a wave period later, the minimum elevation is at the trough when pWL ZK r(g cos a/2u)2. Substituting these pressures into (4) results in the following expressions for wave amplitude: zmax Z g cos a at the crest and 2u2 p g cos a at the trough zmin ZK 2 K1 2u2 The maximum wave height is thus g cos a Hmax Z zmax Kzmin Z p 2 2u (6)

(5)

Pressure pWL denotes the amplitude of the pressure variation at the still-water level. If the pressure at still-water level under
350000 Linear Hogging Sagging

The linear pressures below the design waterline are integrated and extrapolated to obtain pressures between the design waterline and the wave contour. Integration of the resulting pressures then yield the instantaneously acting wave induced loads. These computations are repeated for different wave periods to obtain transfer function corrected for nonlinear effects. Samples of such transfer functions for the waveinduced midship vertical bending moments in head and stern waves of 4.5 m amplitude are shown in Fig. 6; in bow quartering waves (wave heading of 1508) of 2.5 m amplitude and in stern quartering waves (wave heading of 308) of 6.5 m amplitude, in Fig. 7. A statistical analysis based on these transfer functions yielded the long-term hogging and sagging wave-induced
350000 Linear Hogging Sagging

Vertical wave bending moment [kNm]

300000 250000 200000 150000 100000 50000 0 0 1 2 3 4 5

Vertical wave bending moment [kNm]

300000 250000 200000 150000 100000 50000 0

Wavelength / ship length

Wavelength / ship length

Fig. 6. Midship vertical wave bending moment in 4.5 m amplitude head waves (left) and stern waves (right).

302
350000

T.E. Schellin, A.P. de Lucas / Applied Ocean Research 26 (2004) 298308


350000

Vertical wave bending moment [kNm]

Vertical wave bending moment [kNm]

300000 250000 200000 150000 100000 50000 0 0 1 2 3 4

Linear Hogging Sagging

300000 250000 200000 150000 100000 50000 0

Linear Hogging Sagging

Wavelength / ship length

Wavelength / ship length

Fig. 7. Midship vertical wave bending moment in 2.5 m amplitude bow quartering waves (left) and in 6.5 m amplitude stern quartering waves (right).

vertical bending moments. All short-term sea states comprising the wave climate were processed, assuming an equal distribution of wave directionality averaged over all seasons. The commonly accepted PiersonMoskowitz sea spectrum described the spectral density of the seaways, and a cosine squared distribution for the directions of wave energy specied the waves short-crestedness. Absolute values of the resulting midship vertical wave bending moments in hogging and sagging as well as the linearly computed bending moment are plotted in Fig. 8 as functions of the logarithm of their probability of exceedence. As expected, hogging bending moment magnitudes turned out to be less than and sagging bending moment magnitudes greater than linearly computed bending moment magnitudes. Long-term values at a probability level of 2.3!10K8 dened the design values of 267,600 kNm for hogging and K282,500 kNm for sagging. This probability level represents a time at sea of about 6 years during the ships life. 4.3. Rule based loads The rule-based vertical wave-induced bending moments in hogging and sagging were treated separately, i.e., waveinduced bending moments were obtained for each condition. According to classication society rules, the following formulas apply: MWH Z 0:19S=S0 !C !L2 !B !CB MWS ZK 0:11S=S0 !C !L !B !CB C 0:7
2

According to classication society rules, in the sagging condition only an additional vertical wave impact-induced bending moment due to slamming, MSL, had to be added. The procedure to determine MSL takes into account the longitudinal mass distribution, the sectional breadths at the design waterline, the overall length, the service speed, and the rule specied value of 1.87 m/s2 for the vertical design acceleration at the ships forward perpendicular. Inertial loads and slamming loads distributed along the length of the ship were obtained, and lengthwise integration then yielded the slamming-induced vertical bending moment distribution. For the high-speed ferry, the resulting distributions of inertial loads and slamming loads are shown in Fig. 9; the distribution of the slamming-induced bending moment, in Fig. 10. At the ships midship section, this resulted in the value of MSLZK47,603 kNm. The total rule based midship vertical bending moments in hogging, MTH, and in sagging, MTS, were thus obtained as follows:

MTH Z MSW C MWH and MTS Z MSW C MWS C MSL

(9)

For the high-speed ferry, the resulting values were MTHZ276, 777 kNm for the hogging condition and MTSZK302,504 kNm for the sagging condition.
300000

Vertical wave bending moment [kNm]

(7) (8)

250000

Linear Hogging Sagging

200000

where MWH and MWS stand for the vertical wave bending moments in hogging and sagging, respectively. The ships rule length is L, the ships breadth B, and the ships block coefcient CB. Parameters S, S0 and C were specied from the rules. For the high-speed ferry, this resulted in a midship wave bending moment in hogging of MWHZ235,842 kNm and a midship wave bending moment in sagging of MWSZK295,836 kNm. The absolute value of the ratio of sagging bending moment to hogging bending moment thus turned out to be MWS / MWH Z1.25.

150000

100000

50000

0 1,0E-08 1,0E-07 1,0E-06 1,0E-05 1,0E-04 1,0E-03 1,0E-02 1,0E-01 1,0E+00

Probability of exceedance

Fig. 8. Long-term distributions of midship vertical wave bending moment.

T.E. Schellin, A.P. de Lucas / Applied Ocean Research 26 (2004) 298308


300 Inertia loads Pressure loads 200

303

Loading [kN/m]

100 0 100

200

Fig. 9. Longitudinal distribution of inertial loads (light) and slamming loads (dark).

5. Regular design waves The strength of the hull girder was analyzed under the most severe global (design) load conditions. For the high-speed ferry these conditions occur in head waves. Two so-called equivalent regular design waves were specied, one for hogging, with the wave crest located amidships, and one for sagging, with the wave crests near the ships ends. The length of 112.8 m chosen for these equivalent design waves corresponds to linear computations in regular head waves of this length that resulted in the maximum wave bending moment. Wave amplitude dependent hydrodynamic pressures were specied for the dynamically balanced ship positioned in these waves. Balancing was achieved by resolving the motion equations and using the newly found accelerations to adjust the inertial forces. The resulting pressure distributions dened the global wave-induced design loads that, when acting on the hull, subjected the ships midship section to the hogging and sagging vertical wave bending moments. As wave-induced hydrodynamic pressures depended on wave amplitude, so did the resulting wave bending moments depend on wave amplitude. The procedure to determine the amplitude of the two equivalent design waves had to account for this functional relationship. Schematically presented in
10000 0 10 10000 20000 30000 40000 50000

6.5 1.8 2.9 7.6 12.3 17.0 21.7 26.4 31.1 35.8 40.5 45.2 49.9 54.6 59.3 63.9 67.1 71.8 76.5 81.2 85.9 90.6 95.3 99.9 104.6 109.3 114.0 118.7

300

Length coordinate [m]

Fig. 11. Vertical bending moments versus wave amplitude in regular head waves.

Vertical bending moment [kNm]

10

20

30

40

50

60

70

80

90 100 110 120

Fig. 11, this procedure consisted of increasing the wave amplitude until reaching the target value of the design bending moment. For hogging, this target value was obtained by adding the still-water bending moment to the design wave bending moment in hogging; for sagging, by adding the still-water bending moment as well as the slamming-induced bending moment to the design wave bending moment in sagging, see Table 3. This resulted in the design wave amplitudes of 6.12 m in hogging and 5.15 m in sagging. Longitudinal distributions of wave bending moment in hogging and sagging are presented in Fig. 12; longitudinal distributions of the still-water bending moment and the slamming-induced bending moment, in Figs. 9 and 10, respectively. The pressure distributions shown in Fig. 13, one for hogging and one for sagging, represent pressures acting on the hull in regular head waves. These pressures consist of the addition of wave-induced hydrodynamic pressures and still-water hydrostatic pressures. The wave length equals the length of the equivalent regular design wave (112.8 m). For hogging only, the wave amplitude (6.12 m) equals the amplitude of
Table 3 Midship vertical bending moments (kNm) Hogging Sagging K282,500 40,935 K47,603 K289,168

60000

Length coordinate [m]

Fig. 10. Longitudinal distribution of slamming-induced vertical bending moment.

Wave bending moment Still-water bending moment Slamming-induced bending moment Design bending moment

267,600 40,935 0 308,535

304
300000 250000

T.E. Schellin, A.P. de Lucas / Applied Ocean Research 26 (2004) 298308


50000 0

Bending moment [kNm]

Bending moment [kNm]

200000 150000 100000 50000 0 50000 10

50000 100000 150000 200000 250000 300000 10

10

20

30

40

50

60

70

80

90 100 110 120

10

20

30

40

50

60

70

80

90

100 110 120

Length coordinate [m]

Length coordinate [m]

Fig. 12. Distribution of vertical wave bending moment in hogging (left) and sagging (right).

the equivalent regular design wave; however, this is not the case for sagging. Here the wave amplitude (4.40 m) represents a wave that subjects the ships midship section to the design wave bending moment in sagging to which is added only the still-water bending moment; that is, the slamming-induced bending moment is not included. 6. Longitudinal structural strength

applied, they produced unrealistically large local deformations. Therefore, the weight of the ferrys hull structure was introduced as an actual weight curve directly readable by the MAESTRO solver. The resulting longitudinal distributions of weights, vertical shear force, and vertical bending moment for the hogging and sagging conditions are shown in Figs. 14 and 15, respectively. 7. Results and discussion

The longitudinal strength of the hull was assessed by subjecting the ship to the loading conditions specied by the two regular design waves. Wave-induced pressures were distributed as nodal forces acting on the wetted part of the hull. Appropriate input data les were prepared for the nite element model. One le contained nodal forces representing hydrodynamic pressures in the hogging condition, including the static still-water loads. Another le specied nodal forces representing hydrodynamic pressures in the sagging condition, including the static still-water loads and the slamming-induced loads. Inertial forces caused by the ships motion were distributed on nodal masses that idealized the ships longitudinal mass distribution. Nodal forces representing ship weights acted on the center plane at two different heights. These locations did not coincide with nodes of the ships structure. Moreover, it was not possible to interpolate these forces between adjoining nodes, because of the high values of some of these forces. When

The nite element analysis yielded the nominal stress ow (i.e. stresses away from notches and other stress raisers) in the hull girder components of the midship section as shown in Fig. 16 for the hogging condition and in Fig. 17 for the sagging condition. In hogging, the maximum longitudinal stresses amounted to 42.8 N/mm2 in the deck and K65.8 N/mm2 in the bottom; in sagging, the maximum longitudinal stresses turned out to be K44.0 N/mm2 in the deck and 40.8 N/mm2 in the bottom. (Positive values designate tensile stresses; negative values, compressive stresses.) A global view of the corresponding longitudinal stress ow over the deformed shape of the starboard half of the hull is shown in Fig. 18 for the hogging condition and in Fig. 19 for the sagging condition. For comparison, a simple beam theory strength analysis of the ships midship section was performed, based on rule values of the midship vertical bending moments (see Section 4.3 above). The structural members involved comprised all

Fig. 13. Typical pressure distributions in head waves for hogging (left) and sagging (right).

T.E. Schellin, A.P. de Lucas / Applied Ocean Research 26 (2004) 298308

305

Fig. 14. Longitudinal distributions of weights (left), vertical shear force (center), and vertical bending moment for hogging.

Fig. 15. Longitudinal distributions of weights (left), vertical shear force (center), and vertical bending moment for sagging.

longitudinally continuous members that contribute towards structural strength of the primary hull girder in both tension and compression, such as deck plating, shell plating, bottom plating, and longitudinal stiffeners of the hull. These were the items included in the calculation of the section modulus of the midship section. In hogging as well as in sagging, the resulting rule-based magnitudes of maximum longitudinal stresses deviated signicantly from comparable stresses obtained by the nite element analysis. In the deck, stress magnitudes were over 30% and, in the bottom, over 25% greater than comparable stresses obtained form the nite element analysis.

To account for the three-dimensional stress ow in and for the different structural efciencies along the ships midship section, another nite element analysis was carried out, whereby the imposed loads were tuned to the rule values of vertical bending moments; that is, by using wave amplitudes in hogging and sagging to match the rule values of midship vertical bending moment. The resulting magnitudes of maximum longitudinal stresses were signicantly closer to comparable stresses from the previous nite element analysis that was based on equivalent design wave pressures computed by GLPANEL. In hogging, magnitudes of deck stresses turned out to be 9% less than and, in sagging, 13% greater than

Fig. 16. Stress ow in midship section (starboard side) for the hogging condition.

306

T.E. Schellin, A.P. de Lucas / Applied Ocean Research 26 (2004) 298308

Fig. 17. Stress ow in midship section (starboard side) for the sagging condition.

comparable GLPANEL-based stresses. For the bottom structure, the maximum stress magnitude in hogging agreed well with the comparable GLPANEL-based stress; however, in sagging, it exceeded the GLPANEL-based stress by 39%. These results demonstrate that it was important to account for three-dimensional structural effects and structural efciencies when assessing the longitudinal hull girder strength of this high-speed ferry. A summary of the comparative longitudinal stresses is listed in Table 4, together with the corresponding design midship vertical bending moment (VBM). Permissible stresses for the aluminum alloy in the welded condition are 87.5 N/mm2 for plates and 78.7 N/mm2 for stiffeners. These values are based on a minimum guaranteed yield stress of 125 N/mm2 for the plating (Al Mg 4.5 MnA) and 250 N/mm2 for the stiffeners (Al Mg Si 1-T6). For plates as

well as for stiffeners, maximum stress levels were below permissible stresses, thus conrming adequate longitudinal strength of the ship according to classication society rules. An assessment of global hull girder longitudinal structural strength of the high-speed ferry would not have been complete without considering effects of fatigue and buckling. Fatigue, treated in terms of the dynamic stress range (DSR), is critical because of the ships high forward speed, which leads not only to higher wave-induced load magnitudes, but also to an increased number of fatigue load cycles. In addition, the materials needed for a lightweight structure generally are more prone to fatigue damage. The stress ow obtained from the wave-induced (long-term) design bending moments (at a probability level of about 10K8) served to specify the bending moments relevant for

Fig. 18. Longitudinal stress ow in the hull structure for the hogging condition.

T.E. Schellin, A.P. de Lucas / Applied Ocean Research 26 (2004) 298308

307

Fig. 19. Longitudinal stress ow in the hull structure for the sagging condition. Table 4 Comparative summary of hull girder longitudinal stresses Stress analysis method Source of VBM Hogging Design VBM (kNm/103) 309 277 277 Maximum stress in deck (N/mm2) 42.8 55.8 39.0 Maximum stress in bottom (N/mm2) K65.8 K49.2 K65.2 Sagging Design VBM (kNm/103) K289 K302 K302 Maximum stress in deck (N/mm2) K44.0 K61.0 K49.6 Maximum stress in bottom (N/mm2) 40.8 53.8 56.6

FE analysis Simple beam analysis FE analysis

GLPANEL Rules Rules

the assessment of fatigue (at a probability level between 10K3 and 10K4). These so-called fatigue bending moments were in the order of one-half of the wave-induced design bending moments shown in Fig. 8. The corresponding nominal DSR (away from notches and other stress raisers) was therefore calculated from (shogCssag)/2, where shog and ssag represent hogging and sagging stress magnitudes, respectively. The range between hogging and sagging stresses was thus the driving parameter for fatigue. The DSR resulting from the two nite element analyses are summarized in Table 5. They show that a signicant difference in the DSR occurred only at the ships bottom structure. There the DSR was about 12% lower for the GLPANEL-based nite element analysis than for the rule-based nite element analysis. The slope of the cycles to failure relationship (SN curve) in this bottom structural region is close to four. Therefore, in terms of accumulated fatigue damage, this decrease in the DSR would have had an important inuence the ships fatigue life. Buckling considerations of the high-speed ferrys structural design were of greater concern for the upper decks than for the bottom structure because the larger local loads near the ships bottom generally require stronger structures at the bottom than in the upper decks. Thus, for standard panel dimensions, buckling is normally not a problem at the bottom. However,

upper decks generally are made of reduced thickness plating, which is adequate to support local loads. Therefore, the design of decks is normally driven by compressive stresses induced by the dynamic response of the ship in waves.

8. Conclusions The frequency-domain code GLPANEL obtained linear results of hydrodynamic pressures in regular waves of different frequencies and headings. In head, bow, quartering, and stern waves corrections for nonlinear effects increased magnitudes of vertical midship bending moments in sagging and decreased them in hogging. These corrections were well captured by the selected procedure, yielding two separate sets of wave amplitude dependent transfer functions; one set for hogging and another set for sagging. Subsequent stochastic analyses
Table 5 Comparative summary of dynamic stress ranges Stress analysis method FE analysis FE analysis Source of VBM GLPANEL Rules DSR at deck (kN/mm2) 43.4 44.3 DSR at bottom (kN/mm2) 53.3 60.9

308

T.E. Schellin, A.P. de Lucas / Applied Ocean Research 26 (2004) 298308

obtained long-term (design) values as in the standard linear approach. Based on the GLPANEL predictions of wave-induced global loads, a longitudinal strength assessment of the ships structure was carried out by means of a nite element analyses. The resulting stresses were compared with rule based values obtained in two different ways: rst, from a simple beam theory strength investigation of the ships midship section, yielding maximum and minimum nominal stresses in the upper deck and in the bottom structure; second, from another nite element analysis where the imposed loads were tuned to the rule values of vertical bending moments, resulting in stresses that accounted for three-dimensional effects and differing structural efciencies. The rule-based magnitudes of maximum longitudinal stress deviated signicantly (2539%) from comparable stresses obtained by the GLPANEL-based nite element analyses. However, the stresses obtained from the rule-based nite element analysis agreed considerably more favorably with stresses from the GLPANEL-based nite element analysis, thereby demonstrating that it was important to account for three-dimensional structural effects and structural efciencies when assessing the longitudinal hull girder strength of this high-speed ferry. This agreement was more favorable in hogging than in sagging. Magnitudes of nominal compressive stress obtained from the GLPANEL-based nite element analysis were 9% larger in the uppermost deck and 28% smaller in the bottom structure than comparable stresses from the rule-based nite element analysis. Fatigue was treated by examining the nominal dynamic stress range in the deck and the bottom structures of the ships hull. A signicant difference occurred only in the bottom, where the dynamic stress range was 12% less for the GLPANEL-based nite element analysis. Buckling was considered only in stating that it was more likely to occur in the uppermost deck than at the bottom. This is because, generally, the upper decks are just thick enough to support local loads. Therefore, the design of decks is normally driven by compressive stresses induced by the dynamic response of the ship in waves.

Acknowledgements The presented work was performed under research project BE-4406, Advanced Methods to Predict Wave-Induced Loads for High-Speed Ships (WAVELOADS). This project was partially funded by the Commission of the European Community under the BRITE/EURAM program, contract BRPR-CT 97-0580. References
[1] Chen X-B, Noblesse F. Super green functions Proceedings of the 22nd symposium on naval hydrodynamics, Washington, DC. Washington, DC: National academy Press; 1998. [2] Schellin TE, Beiersdorf C, Chen X-B, Fonseca N, Guedes Soares C, Maron Loureiro A, Papanikolaou AD, Perez de Lucas A, Ponce Gomez J. Numerical and experimental investigation to evaluate wave-induced global design loads for fast ships. Trans SNAME 2003;111:43761. [3] Ostergaard C, Schellin TE. Development of an hydrodynamic panel method for practical analysis of ships in a seaway. Trans Schiffbautechnische Gesellschaft 1995;89:56176 (in German). [4] Papanikolaou AD, Schellin TE. A three-dimensional panel method for motions and loads of ships with forward speed. Ship Techn. Res. 1992;39: 14756. [5] Hachmann D. Calculation of Pressures on a Ships Hull in Waves. Ship Techn Res 1991;38:11132. [6] Schellin TE, Beiersdorf C, Chen X-B, Maron A. Comparative frequencydomain seakeeping analysis of a fast monohull in regular head waves Proceedings of the 22nd international conference on offshore mechanics & Arctic engineering, OMAE 28502, Oslo 2002. [7] Guedes Soares C. Long-term distribution of nonlinear wave-induced vertical bending moments. Marine Struct 1993;6:47583. [8] Guedes Soares C, Schellin TE. Long-term distribution of nonlinear vertical bending moments on a containership. Marine Struct 1996;9: 33352. [9] Torhaug R, Winterstein SR, Braathen A. Nonlinear ship loads: stachastic models for extreme response. J Ship Res 1998;43:4655. [10] EEIG UNITAS. Rules for classication and construction, Iship technology, part 1seagoing ships, chapter 5 - High Speed Craft. Hamburg: Germanischer Lloyd; 1998. [11] BMT. Global wave statistics;user manual PC GWS V2.10. Middlesex: BMT Fluid Mechanics Ltd; 1990. [12] Fonseca N, Guedes Soares C. Time-domain analysis of large-amplitude vertical motions and wave loads. Journal of Shipe Research 1998;42(2): 10013.

Potrebbero piacerti anche