You are on page 1of 9

Nonlinear Analysis of Heavy- Lift Barge Roll Motion

Allen H. Magnuson, Ph. D., P. E.


Consulting Engineer 935 E. Magnolia St. Angleton, Texas 77515 USA
Abstract - Two types of nonlinearity pertinent to heavy-lift barge roll motion were treated: quadratic damping and nonlinear roll restoring moment . The Ritz method based on variational calculus was applied to solve the nonlinear equation of motion. Extensive large amplitude roll motion wave tank data on a series of seven barge configurations was reanalyzed using this method. A comparison study of roll responses was conducted for three key barge design parameters: draft, length-beam ratio, and roll period. Quadratic ( || ) damping was used to represent effects of wave impact, deck edge immersion, and bilge emergence. Nonlinear restoring moments were introduced by modifying GM to account for deck cargo immersion and green water on deck at large angles of roll. Characteristic features of the nonlinear responses included shift of the natural frequency with increasing roll angle and localized instabilities in roll angle and phase angle.

on deck. Deck permeability (fraction of deck flooded) decreases the GM and natural frequency. In addition, deck cargo overhanging barge sides increases the wetted deck area at larger roll angles, increasing GM and natural frequency. Both nonlinear effects were analyzed using example problems. Roll Angle Data Analysis The model test data in [1] was reanalyzed. Plots of peak values of roll angles vs. wave height for the seven NDA barge series in regular beam seas are shown in Figure 1. The wave heights have been scaled up to full scale. This data indicates that the peak roll angles tend to level off as wave height increases. The nonlinear behavior starts at about 1.5 m. This marked fall-off in roll angle with wave height is an indication of strong nonlinear behavior. The default NDA 20 degree roll angle guideline value [3] representing the safe operating limit is also shown. (This limit is applicable to barges with a ten second wave period). Note that most of the roll angle data is well above the 20 degree NDA guideline limit. This makes the data ideal for nonlinear studies.

INTRODUCTION Roll motions of heavily loaded barge-type hull forms, either self-propelled or under tow, in ocean transit can be problematic in heavy seaways because of the vessels high VCG, low freeboard, and shallow draft. Existing commercially available six degree of freedom ship motions computer programs are basically linear with, if anything, weak roll nonlinear terms that under-predict roll damping. As a result, they grossly over-predict roll motions, making them virtually useless for predicting roll motions for barges. Extensive model-scale regular wave beam seas roll motion data on a barge series was acquired in an earlier JIP effort headed up by Noble Denton Associates (now GL Noble Denton) [1]. This data was re-analyzed in the present study Nonlinear square-law v|v| Morisons equation-type [2] damping was used because it is associated with stagnation pressure. (Stagnation pressure is: pd = V2.) At large roll angles wave impact occurs on the barge deck, sides and bottom. The impacts are a result of deck edge immersion, and/or bilge emergence combined with wave interactions. Nonlinear Cargo Immersion Study Nonlinear roll stiffness is associated with hydrostatic or metacentric stability, and is governed by the metacentric height GM. In large amplitude roll motion the GM is modified because of green (relatively deep, non-spray) water
978-1-4244-4333-8/10/$25.00 2010 IEEE

Figure 1. Roll Angle vs. Wave Height for All Cases

This nonlinear behavior at large roll amplitude is caused by a combination of deck edge immersion, bilge emergence and wave impact on barge sides, deck, and green water on deck. The nonlinear behavior is a result of the low freeboard and shallow draft of barges as compared to cargo vessels.

One can see that for safe operation in severe sea states, there is an urgent need for a simple but accurate barge roll motions prediction tool that can model both nonlinear roll stiffness and nonlinear damping. Scale models of all seven barge cases in Table 1 were tested by NDA [1] in a wave tank in regular (constant frequency) beam seas for various wave heights and frequencies. The barge series was designed to determine the motion characteristics as a function of variation of three key parameters: Length to beam ratio (L/B), draft to beam (t/B), and roll period Tn. The Case 3 barge is the Standard Barge used as the reference point for all three variations. DESCRIPTION OF BARGE CASES The barge series has three basic parametric variations: Length: (Cases 1, 2, and 3) Draft: (Cases 3, 4 and 5), and Roll period T: (Cases 3, 6 and 7). Key barge characteristics are given in Table 1. Figure 2 from [1] shows key full scale dimensions of the barge.

Figure 2. Key Barge Dimensions [1]

UNCOUPLED ROLL NONLINEAR EQUATION OF MOTION Uncoupled nonlinear roll motion in unidirectional sinusoidal beam-seas regular waves is treated. The uncoupled roll motion equation with nonlinear expressions for both damping and stiffness can be written as [4]: d2/dt2 +2 h(d/dt) +2 f () = p sin( t) (1)

TABLE 1 BARGE CHARACTERISTICS Case Var. L/B Draft (m) Nom. Roll Pd. (sec) 10 10 10 10 10 6 14 KG (m) rxx (m) Displ. (metric ton) 9050 4817 6581 11173 5086 6581 6581

Where: f () = Normalized nonlinear roll stiffness factor, h(d/dt) = Normalized nonlinear roll damping function, d2/dt2 = Roll acceleration, d/dt = Roll rate, = Roll angle, = Nondimensional roll damping ratio, p = Normalized roll moment, and = wave excitation frequency. The corresponding frequency-domain nonlinear roll motion equation corresponding to (1) can be written using ship dynamics notation as [5]: [-2(Ixx + A44) + i b44 ()+C44]() = F4( ) (2)

1 2 3 4 5 6 7

L/B L/B Std. Barge Deep Draft Light Draft Short Roll Pd. Long Roll Pd.

4.5 2.5 3.33 3.33 3.33 3.33 3.33

2.74 2.74 2.74 4.57 2.13 2.74 2.74

9.76 9.76 9.76 6.4 11.59 4.88 12.2

15.79 15.79 15.79 13.08 18.11 8.48 22.01

Where = Wave frequency Ixx = Roll moment of inertia, A44 = Hydrodynamic inertia roll moment, b44() = Roll damping coefficient, C44( ) = Hydrostatic roll moment coefficient, F4() = Wave roll moment n2 = C44( )/(Ixx + A44) = Natural frequency squared b44() = ()[ 2 C44( )(Ixx + A44)] The nonlinear roll stiffness factor f () in (1) can be written in naval architectural terms as follows [6]: f () = GZ()/GM Where GZ() = Righting moment arm and GM = Metacentric height. (3)

The stiffness nonlinearity GZ for the wall-sided ship (vertical sides) can be written as [6]: GZ = GM h(), where h() = d1+d32 + d5 4 (4) Values for the coefficients d1, d2, and d3 are given in [6]. The wall-sided correction for GZ is too small to be significant for barges because of their low freeboard. The normalized time-domain roll damping function h(d/dt) was taken as the velocity-squared (quadratic) roll rate damping given as [4]: h (d/dt) = 2 d/dt |d/dt| , (5)

Figures 3 and 4 show the roll amplitudes and phase angles, respectively for the Case 3 barge with a 3 m wave height. The test data is shown as points, while the 3 m computed responses are the continuous lines. One sees quite good agreement for the roll angles. The phase angle data shows considerable scatter, but with good overall agreement. Case 3 was the only case where phase angle data was taken.

where 2 is the wave height and frequency dependent damping coefficient. The frequency domain total roll damping coefficient b44 is: b44() = 2 D 2 n (Ixx +A44) 8/(3 ) 0 (6)

where linear damping D = = b44L /[2 c44(Ixx +A44)] D = 0.02 (2%) was used throughout ROLL FREQUENCY RESPONSE FUNCTIONS CURVE-FIT TO WAVE TANK MODEL TEST DATA Roll frequency response functions (magnitudes and phase angles) were computed using the frequency domain uncoupled roll equation (2) for all seven cases for all wave heights. The frequency response functions were curve-fit to the NDA model test data presented in [1]. The curve-fitting was performed by varying the quadratic damping coefficient 2 and the wave moment F4 as needed to obtain good agreement with the test data for each wave height. Roll moment (F4( )) computed in [1] used Franks close-fit (linear) procedure [7]. The roll amplitude curves initially produced peaks that were too narrow. The curves were widened by increasing F4. Good agreement was achieved for all cases by first curve-fitting the width of the frequency response function by adjusting F4 , then adjusting the roll peak using the damping function 2. This data was then used to obtain a set of curve-fit damping coefficients evaluated at n as a function of wave height for each configuration. This data was then plotted up to display visually the relationships between the roll angles and damping coefficients and the key design parameters: L/B, t/B, and roll period Tn. This data can be used to predict roll motions for any barge configuration with design parameters within the limits of the NDA series [1] by using scaling laws and interpolation. Figure 3. Roll Angle vs. Frequency for Case 3

Figure 4. Phase Angle vs. Frequency for Case 3 ROLL ANGLE AND DAMPING VS. WAVE HEIGHT: ALL THREE VARIATIONS Roll angle and damping coefficient data is shown below for all seven barge cases. The data is presented in three groups consisting of three sets of three load cases for variations in: 1. Draft (t), 2. Roll Period (T), and 3. Length/Beam ratio (L/B). The first group to be analyzed is:

BARGE DRAFT EFFECT ON ROLL ANGLE AND DAMPING

Table two shows the barge case listing and draft data.
Table 2: Draft Variations Case No. 3 4 5 Draft designation Design draft Deep draft Shallow draft Draft (t) (m) 2.74 4.57 2.13 Ratio t/B (%) 10 16.68 7.77

NDA Guidelines [3] specifies limiting values for barge drafts as a function of barge depth (D) as follows: 0.35D < t < 0.6 D The barges all have depths: D = 6.1 m. This translates into the following range of acceptable drafts for the barge series: 2.135 m < t < 3.66 m Comparing these limits to the drafts in Table 2 shows that Barge 4 (deep draft) is considerably overloaded. Barge 5 (shallow draft) is slightly under-loaded. Barge 3 (standard barge draft) has acceptable loading. Figure 5 shows a plot of roll angle vs. wave height for cases 3, 4 and 5. The Case 3 and 5 data show the usual falloff on roll angle vs, wave height. Case 3 (design) shows the highest roll angles, Case 5 (shallow draft) shows a lower roll angle curve parallel to the Case 3 data, probably due to bottom slamming. The Case 4 (deep draft case) shows an abrupt downward break at 1.5 m wave height due to severe wave impact and green water on deck. It is informative to compare these roll angle results to the NDA default motion criteria in section 7.9 of [3]. This data gives maximum acceptable values for roll and pitch angles and heave accelerations for specified roll periods and vessel types. The 20 degree roll angle limit is plotted in Figure 3. The limiting wave heights for each Case are shown below. Case 3: 1.5 m (extrapolated) Case 5: 2.5 m Case 4: 5 m Figure 6 (damping vs. wave height) shows that damping and roll angles have opposite trends: low roll angles correspond to high damping and vice versa. This is because damping by definition damps out motion. The highly nonlinear behavior in Cases 4 and 5 will be shown to be primarily due to deck edge submergence and bilge emergence augmented by wave impact.

Figure 5. Draft Variation: Roll Angle vs. Wave Height for Cases 3, 4, and 5.

Figure 6. Draft Variation: Damping at n vs. Wave Height for Cases 3, 4, and 5

The roll damping plotted in Figure 6 exhibits nonlinear behavior, particularly for Case 4, the deep draft case. The Case 4 roll damping line abruptly branches off the Case 5 line at 1.5 m wave height. This break is primarily a result of incoming waves breaking or impacting the essentially swamped deck. Similarly, the damping increase of the shallow draft Case 5 relative to Case 3 was evidently due to bottom pounding associated with bilge emergence. This can be verified by examining the submergence and emergence data in Table 4 [1].
Table 4 Heel Angles of (Calm Water) Deck Edge Submergence and Bilge Emergence Barge Case Freeboard (m) Draft (m) Heel Angle of Deck Edge Submergence (deg) 14 6 16 Heel Angle of Bilge Emergence (deg) 11 18 8

3 4 5

3.36 1.53 3.97

2.74 4.57 2.13

Barge case 4 (deepest draft) with a freeboard of only 1.53 m. (full scale) experienced the most green water on deck in the model tests, and also experienced wave impact on a horizontal plate projecting over the sides.[1] The wave impact evidently produced considerable inertial (square-law) roll damping. Deck edge calm water submergence started at only 6 degrees of roll. One can see from Table 4 that the roll angles for all three cases in Figure 4 are well within the deck edge submergence and bilge emergence range. This indicates that the (nonlinear) roll damping was increased by wave impact (green water on the deck, deck cargo, and impact on the bottom), resulting in highly nonlinear damping, which reduced the roll motion. This is confirmed by the correlation of the reduced roll angles with values of the damping coefficients at the higher angles seen in Figure 5.
EFFECT OF ROLL PERIOD VARIATION ON ROLL ANGLES AND DAMPING COEFFICIENTS

Figure 7. Roll Damping at n for Three Roll Periods

Figures 6 and 7 show plots of roll angle and damping vs. wave height, respectively. Roll periods were: T = 10 sec. (case 3) T = 6 sec. (case 6), and T = 14 sec., (case 14). From Figure 6 one sees that Case 3 (T=10) had the largest roll angles, followed by Case 7 (T = 14 sec.). Lowest roll angle was for T = 6 (case 6). From Figure 7 one sees that Case 3 (T=10) had the highest damping, followed by Case 7 (T = 14 sec.). Lowest damping was for T = 6 (case 6). The roll period variation for both roll and damping was the same: highest Case 3, then Case 7, and finally Case 3.

Table 5a tabulates 3 m. wave height data for roll and damping (b44). Also shown are wavelength () and wavelength/beam (/B) ratio. Case 6 has a (/B) of 2.05. This low value of (/B) explains the low value of the roll angle for L/B = 6. The short wave length reduces the roll angle and damping, apparently because the wave roll moment is reduced. Table 5a. Roll Period Variation Analysis Results
3 m W.H. Case 3 6 7 nom T (sec) 10 6 14 Roll (deg) 26 21.5 23.7 3 m W.H. b44 2.64E+05 1.10E+05 2.00E+05 WL

156.06 56.18 305.88

/B
5.70 2.05 11.16

Table 5b shows how the roll period was varied by varying the radius of gyration (rxx), and the moment of inertia (Ixx). Note that the metacentric height (GM) was kept approximately the same.
Table 5b. Variable Roll Period Inertial and Stability Data
Rad of Gyr Case 3 6 7 rxx (m) 15.8 8.5 22.0 (rxx/B)% 57.66 31.02 80.29 (I xx/m)2 (rxx/B)2 % 33.2 9.6 64.5 KG 9.8 4.9 12.2 KG/B % 35.6 17.8 44.5 GM 15.1 20.0 12.7

Note also that the roll angles and damping do not have an inverse relationship like they did for the draft comparison. For the roll period variation the roll angles and dampings have the same rankings. Evidently the roll period variations do not significantly change the damping qualitatively relative to the roll angles.
LENGTH/BEAM RATIO (L/B) EFFECT ON BARGE MOTIONS AND DAMPING Figure 6. Roll Angle for Three Roll Periods

Figures 8 and 9 show variations of roll angle and damping with Length-Beam ratio. Figure 8 shows that Case 3 (design case) had the highest roll angles, next was Case 2 (L/B = 2.5). Lowest roll angle of 20 deg was for Case 1: L/B = 4.5. Note that the case 2 and 3 roll angles exceeded the NDA 20

degree guidelines. Case 1 roll at 3 m was right on the guideline.

Table 6. Barge Response Variation With L/B Ratio at 3 m Wave Height


Case 1 3 2 L/B 4.50 3.33 2.50 Rel. Length Longest Mid-range Shortest L 123.3 91.2 68.5 Roll (deg) 19.70 26.00 23.70 Damping 514,900 263,700 214,000 Dmp'g/L 4,176 2,890 3,124

SCALING OF BARGE DATA

Figure 8. Roll Angle vs. Wave Height: Three L/B Ratios

The barge data shown above can be applied to geometrically similar vessels, i. e., it can be scaled. The simplest scaling is direct geometrical scaling where every dimension is multiplied by the same factor. The geometrical scaling is governed by the scale ratio . For example for length: = Ls/L0, where Ls = scaled length and L0 = original length. Wave height scales the same way: WHs = WH0 . One must also scale roll natural frequency n. One can use the nondimensional frequency to scale, where = ( B/g). One can also scale on length, draft and roll period by interpolating using the variational data given above. SOLUTION OF THE NONLINEAR EQUATION OF MOTION WITH NON-LINEARITIES IN ROLL RESTORING MOMENT This paper applies, apparently for the first time for a marine application, the Ritz method developed earlier for nonlinear vibration analysis [4]. This method treats nonlinearities for both nonlinear power-law damping and general restoring forces. Expressions for the barge roll amplitudes and phase angles were derived using this method. We wish to solve the nonlinear differential equation (1) with normalized nonlinear damping factors h(d/dt) and general restoring factor f (). Following [4], a solution is assumed in the following form: d2/dt2 = A cos(t - ). (7)

Figure 9 shows variation of damping with L/B. Highest damping was for Case 1 (L/B = 4.5: longest barge). Next was Case 3: (L/B = 3.33). Lowest was Case 2: L/B = 2.5.

Figure 9. Roll Damping n vs. Wave Height: Three L/B Ratios

Table 6 shows a tabulation of barge data for Cases 1, 3, and 2. One would expect the damping per unit length to be approximately the same, as the fluid motion should be largely two dimensional, at least for the higher L/B ratios. The last column shows damping per unit length. One sees that the variation of roll angles is moderate, as one would not expect the roll angle to be strongly affected by length. Barge 1, the longest barge, has the smallest roll angle and the highest damping per unit length, as one would expect. Case 3, the design reference, has the highest roll angle and the lowest damping per unit length, again as one would expect. However, Case 1, the longest barge, should have the lowest damping per unit length, as end effects should be minimized, but instead, it has the highest. The reason for this is not evident.

This solution is substituted into (1) and the Ritz conditions are applied. The result is a definite integral F(A), which can be integrated. After considerable algebraic manipulation, one gets two independent equations for the two unknowns A, and . These are then solved to get expressions for the responses , , and frequency ratio . In the present investigation calculations of roll angles and phase angles vs. frequency were performed using design data for all seven cases from [1]. Figures 2 and 3 for Case 3 were calculated in this way. These calculations were performed using square-law damping, taking roll stiffness as a constant. Nonlinear Roll Stiffness Studies Two sets of calculations with nonlinear stiffness were performed. The first was on the effects of cargo immersion, based on NDA Guidelines, sec. 7.6 [3]. The guidelines suggested that: The effect of cargo immersion in increasing

GM and hence reducing natural roll period as well as increasing damping should be considered in motion response analyses. Deck cargo immersion will increase effective GM, resulting in a nonlinear stiffening effect seen as an increase in the natural roll frequency as the roll angle increases. Nonlinear Deck Cargo Immersion Study A study was performed for the Case 3 barge with the 3 m. wave height and 25 degree roll angle. Overhanging deck cargo representing a 50 m square production topsides unit was placed amidships. Surface permeability of 30% was used to account for the open topsides structure. The original GM was 15.1 m. With the deck cargo the GM increased to 21.9 m. The natural frequency increased from 0.625 rad/sec to 0.753. Roll period decreased from 10 sec to 8.34 sec. Frequency response plots taking into account the increase in stiffness are shown in Figures 10 and 11 for roll angle and phase angle, respectively. The backbone (BB) curve [8], shown as the dashed black line in Figure 10 defines the dependence of the frequency ratio / on the roll angle in undamped free oscillations. The stiffening backbone curve inclines considerably to the right. The peak frequency ratio is now / = 1.2, vs. 1.0 for the linear case. Note that the roll angle peak curve is essentially vertical on the right hand side. This means if the wave frequency increases past the peak the roll angle will suddenly oscillate between about 13 to 20 degrees at the wave frequency of / = 1.2. This is unstable behavior. Also note the crossing locus curve shown as the green dashed line. This curve crosses the backbone curve to indicate the precise value for the roll angle peak. Also note that the barge at sea in a swell in resonance with a period of 8.3 sec. will be rolling erratically because of the instability mentioned above. Also note that the roll period will be lower than the undamped period of 10 sec.

the roll peak in Figure 10. Here the phase angle can range erratically from 90 to 150 degrees at the resonant frequency of / = 1.2. This is another indication of motion unstability at this frequency. Note also the sudden curve shape break on the right hand side as opposed to the smooth transition seen in linear roll motion. The erratic oscillation at resonange is called the jump phenomenon in nonlinear vibration theory[4].

Figure 11. Effect of Deck Cargo Immersion on Roll Phase Angle

Nonlinear Deck Immersion Study Results of the study of the effects of deck immersion are described below. Deck immersion can be modeled approximately by reducing the GM to account for free water on deck. The free water is caused by a combination of oscillatory deck edge submergence and breaking waves. The result may be referred to as green water on deck. The free water has the effect of reducing the hydrostatic stability and raising the VCG. At higher angles of roll, this effect can be modeled by reducing the waterplane area moment of inertia Ixx. The effective Ixx can be reduced by applying a surface permeability factor. Here the permeability was taken as 70 percent (water coverage) which is appropriate for a lightly loaded barge deck rolling in a severe sea state. The uncorrected GM was 15.1 m, while the partially immersed wet decks GM was 8.8 m., a considerable reduction. The reduction in stability increased the roll period from 10 sec to 13.2 sec. The resulting frequency ratio ( / n) was 0.764. The frequency plots for roll angle and phase angle are shown in Figures 12 and 13. The softening (BB) line in Figure 12 curves considerably to the left. The magnitude peak plot (Figure 12) shows a concavity on the left hand side of the peak. This will result in a sudden downward jump in roll angle from 25 to about 10 degrees when the wave frequency decreases past the peak of (/ n) = 0.764. Similarly, the phase angle data in Figure 13 indicates a nearly vertical line from 75 to about 10 degrees corresponding to the jump in Figure 12. This results in unstable roll angle motion or jump phenomenon behavior similar to the cargo immersion instability. As mentioned before, this type of nonlinear response is well known in the mechanical vibration field [4].

Figure 10. Effect of Deck Cargo Immersion on Roll Angle

Figure 11 shows the phase angle for the deck cargo immersion load case. Note the steep slope for the phase angles above 90 deg. This corresponds to the steep slope in

noted that his kind of nonlinear response is well known in the mechanical vibration field [4]. CONCLUSIONS The Ritz variational calculus-based method was applied (apparently for the first time in the marine field) to compute barge large amplitude nonlinear roll motions. A quadratic damping model was used to analyze roll motions using data from the DNV barge series [1]. Two investigations of the effects of combined quadratic damping and roll stiffness nonlinearity were conducted: one for a stiffening system and one for a softening system. The results showed that a roll motion instability called the jump phenomenon [4] was introduced by the roll stiffness nonlinearity. The method is relatively easy to use, and provides useful information not otherwise available on barge motions in severe sea states. RECOMMENDATIONS This frequency-domain nonlinear roll treatment should be extended to good effect to analyze extreme responses for (conventional) ships. A six degree of freedom frequency domain simulation including these nonlinear roll features should also be implemented. In addition, the procedure should be extended to treat barge coupled roll-sway motion. Roll and sway are closely coupled and such coupling could, among other things, enable accurate modeling of the vertical shift in the roll center, which is frequency dependent. Because of the nonlinearities, simple superposition of the frequency domain motion solutions given here cannot readily be used to compute ship responses to random waves the way one can for linear frequency domain representations. Instead, nonlinear perturbation techniques can and should be applied. (See, e. g. [8].) ACKNOWLEDGEMENTS The author is grateful to and wishes to thank his former supervisors Jorge Hernandez and Dana Kelley at Bastion Technologies, Houston for partially supporting this work. He also thanks S. W. Tseng, also from Bastion, for his technical advice. REFERENCES [1] Noble Denton & Associates Limited (Now GL Noble Denton), Barge Motions Research Project Summary Report, Report No. L12140/NDA/JBW, July 1984. [2] Morison, J. R., OBrien, M. P., Johnson, J. W., & Schaaf, S. A. The force exerted by surface waves on piles, Pet. Trans., Vol. 189, pp. 149-54, 1950. [3] Noble Denton, General Guidelines for Marine Transportations No. 0030/NDI Rev 2, 2005.

Figure 12. Effect of Deck Immersion on Roll Angle

Figure 13. Effect of Deck Immersion on Roll Phase Angle

SUMMARY The Ritz method using calculus of variations has been applied to treat nonlinearities in the analysis of the barge underdamped time-harmonic roll motion problem. Two types of nonlinearities were treated: nonlinear roll damping and nonlinear roll stiffness. In the first part of the investigation nonlinear roll damping was used to back-calculate frequency response functions for the seven NDA[1] barge cases by curve-fitting using the quadratic damping coefficients quadratic b44. The results were used to determine the effects of variation of draft, roll period, and length-beam ratio on roll angle and damping. In the second part both square-law and roll stiffness nonlinearities were applied to investigate deck cargo immersion and deck immersion. It was shown that motions with the roll stiffness nonlinearity exhibit responses quite different from linear time-harmonic response. The amplitude peaks are no longer symmetrical. Instead, the peaks shift to the left or right depending on whether the roll stiffness decreases or increases with amplitude. It was shown that the GM-related nonlinearities for both cases introduced an instability in roll angle and phase due to the lateral shift of the peak frequency. It was

[4] Shock and Vibration Handbook, eds.: C. M. Harris and C. E. Crede, Second Edition, Ch. 4, pp. 4.4-4.21, 1976. [5] J. N. Newman, Marine Hydrodynamics, The MIT Press, Cambridge, Massachusetts, 1977. p. 308. [6] Price & Bishop,Probabilistic Theory of Ship Dynamics London: Chapman and Hall Ltd., 1974. [7] Frank, W., Oscillation of cylinders in or below the free surface of deep fluids, Report 2375, Naval Ship Research and Development Center, Washington D. C., 1967.

[8] Belenky, V. L. & N. B. Sevastianov Stability and Safety of Ships, eds.: R. Bhattacharyya & M. E. McCormick, The Society of Naval Architects and Marine Engineers, Jersey City, N. J., 2007. [9] J. B. Roberts and P. D. Spanos, Random vibration and statistical linearization, Dover Publications, Inc., Mineola, New York, 1999.

You might also like