Prediction of the effects of fast passing vessels on moored vessels

Date 2013 ^ Auttior Pinkster, J.A. and J,A. Keuning

Address Delft University of Technology

Ship Hydromechanics and Structures Laboratory Mekelweg 2, 2628 CD Delft

Delft University of Teclinology

TUDelft

on O c e a n , Offshore and Arctic E n g i n e e r i n g , O M A E 2 0 1 3 , J u n e 9 - 1 4 , 2 0 1 3 , Nantes, F r a n c e , Paper 1 0 6 3 1 .

Proceedings of the ASME 2013 32nd International Conference on Ocean, Offshore and Arctic Engineering OMAE2013 June 9-14, 2013, Nantes, France

OMAE2013-10631

PREDICTION OF THE E F F E C T S OF FAST PASSING V E S S E L S ON MOORED

V E S S E L S

J.A. Pinkster

Pinkster Marine Hydrodynamics BV Rotterdam, Ttie Netherlands

J.A. Keuning

Delft University of Technology Delft, The Netherlands

ABSTRACT

An analysis method to determine the behaviour of a moored ship due to wash waves created by a passing fast vessel has been developed . The method can be used for quantitative assessment o f the motions of the moored vessels due to wash waves.

In this paper a short review of the main elements of the computational method is given and comparisons made with results of model tests carried out at M A R I N with the concept of a fast passenger feiTy and a captive barge representing a moored vessel.

Besides having the possibiUty to analyse, i n the design stage o f a fast vessel, the effects on other waterway users, criteria are needed to determine acceptable limits for the motions and accelerations of the moored vessel as caused by the passing vessel. In this paper motion criteria developed based on a comprehensive field study carried out in the Netherlands using the experiences of crew members of moored barges are compared with results of computations of barge motions.

INTRODUCTION

Fast ships create water motions which appear i n the form of waves. Relative to the passing ship itself, the wave pattern is stationary but for moored ships the waves represent a transient , dynamic disturbance which may cause unexpected motions hampeiing loading/discharging operations and , in extreme cases, can lead to personal injury or breakage of mooring lines. With the advent of more , larger and faster vessels on inland waterways passing ship effects are on the increase with corresponding increases i n the material damage, loss of time and personal injuries. A recent survey carried out by

the Netherlands ministry of transport and waterways among the skippers of inland vessels revealed that some 40 % of the damage sustained to mooring equipment were due to passing ship effects. The same study has resulted i n a re-assessment of the design loads of mooring equipment o f such vessels to reduce the number of serious incidents.

Many of the incidents recorded i n the past have focussed on the wave-induced motions o f the moored vessel as the cause of personal injury and damage to mooring systems. In the Netherlands a comprehensive study of the effect of passing vessels on the subjective assessment o f the ensuing motions of a moored inland waterway vessel by a number of professional inland waterway skippers on board the vessel was carried out. See Bos et al, 2007,Ref [1] and Luth et al, 2009, R e f [2]. This smdy, ordered by the Province of South Holland (The Netherlands) which oversees fast waterborne passenger transport i n the province, has lead to new criteria for the allowable motions of moored inland waterway vessels due to passing ships. I n the paper attention w i l l be paid to this study and the developed motion criteria.

In order to allow assessment o f a new design o f fast inland water way vessels, the criteria can be applied i n the design stage using numerical procedures f o r the assessment of the wave field due to a passing vessel and a motion analysis.

E F F E C T S OF WASH WAVES OF FAST CRAFT

We base the development of the prediction method f o r the disturbance due to wash waves on the input due to a so-called 'wave-cut' of a wash wake pattem taken a short distance f r o m the centreline o f the passing fast vessel.

Short distance in this context means , close to the vessel but outside the region of strong non-linearities (steep or breaking waves) so that linear wave theory may be applied to the waves progressing away f r o m the vessel beyond the wave cut. Based on this relatively short wave record we propagate the wash waves to the surrounding port geometry and moored vessels using linear potential wave theory. We have chosen linear potential flow to propagate the wash waves pardy based on the findings o f Whittaker et al (2001), R e f [3]. with respect to fuU scale measurements of the wake wash of fast ferries. He concluded from analysis of the measurements that the propagation of the wash waves conformed well with predictions based on linear dispersion.

Basic elements of the computational method

A wave cut is the transient time record of the wash wave elevation i n a fixed point taken as the vessel passes close by on a straight course at constant speed. The record can be obtained based on computations , model tests or f u l l scale measurements. This is an interesting aspect of the approach since it introduces a degree of flexibility i.e. a f u l l scale measurement of a wash wave can be based on a single wave-rider buoy measurement for an existing vessel. I f it concerns a new design, computations may suffice i f the hull f o r m is amenable to a numerical analysis and i f not a short model test program with the new design can be carried out to measure the wave-cut. A example of a computed wave field using Raven's code RAPED (Raven ,1996, R e f [4]) is shown i n Fig. 1.

Figure 1 : Wash wave field computed by RAPID (Raven,1996)

A n example of a wave-cut taken from a model test is shown i n Fig. 4.

The procedure to obtain the pressures and velocities induced by the wash waves on the port geometiy and the moored vessels involves extending the duration of the

wave-cut by padding the record with zeros to the expected duration of the simulation. This is related to the propagation time of the waves to the moored vessels and the time for the vessel response to reduce to near zero. By Fourier analysis of the thus lengthened, measured or computed wave-cut the frequency-dependent complex amplitude of the wave-cut is obtained :

f , „ ( « ) = F ( f „ , ( 0 } (1) Each wave frequency component has its own, unique

wave direction. We compute the wave direction of each frequency component by assuming the vessel is travelling in the direction of the positive -axis at constant speed U . The wave direction of each component can be computed knowing the speed of the passing vessel U and the water depth / ; . This follows f r o m the requirement that the projection of the sailing speed vector in the direction of wave propagation be equal to the phase speed of the wave. See Fig. 2.

Figure 2 : Direction of a frequency component of the wash waves

This leads to :

a(a) = a r c c o s ( ^ ) (2)

In which the frequency-dependent phase speed c{w) follows f r o m :

cm=^ (3) k

The wave number k follows f r o m the solution of the dispersion relationship :

üY =kgtai\hkh (4)

Using hnear potential theory we can compute, f o r each wave frequency, the undisturbed wave pressure and velocity of the wash wave components on the grid o f the port model and the panel models o f the moored ships with co-ordinates , X2, X3 i n an earth-fixed system of axes. These quantities are derived f r o m the velocity potential which is given by :

8 cosh kh

This data can be used as input for frequency-domain linear diffraction computations. I n our case we use the program DELFRAC. See Pinkster, 1996,Ref [5], Pinkster and Naaijen, 2003, Ref [6]. Results of Eqn. (5) replace the usual long-crested frequency-domain wave input with unit wave amplitude , zero phase angle and fixed direction. I n our case each wave frequency component in a given wave cut has its own direction, amplitude and phase angle.

After solving the frequency-domain diffraction problem for wave forces , wave frequency motions and wave elevations, time-domain results for forces, motions and wave elevations are obtained by inverse Fourier transform methods. Besides these quantities, it is also possible to compute the mean and low-frequency second order wave forces acting on the moored vessel. See Pinkster and Naaijen, 2003, Ref [6].These are left out of consideration i n this paper.

Application to a Fast Ferry passing the captive model of a rectangular barge.

The computations concern a fast passenger ferry with a displacement of approximately 90 tons saiUng at 34 kn in 10.0 m water depth. The model tests were carried out at scale 1:25 i n the Shallow Water Basin at M A R I N , Wageningen. This basin measures 210 m x 15.8 m x 1 m(max). For the tests the water depth amounted to 0.40 m . The basins sides are vertical, smooth and rigid thus assuring f u l l refiection of wash waves generated by the passing vessel.

The set-up for the tests is shown i n Fig. 3 using f u l l scale data. This Figure shows the fast ferry (schematic) sailing on the basin centreline and an earth-fixed array of wave probes along with the captive model o f a 80 m x 10 m x 2.5 m barge .

The wave-cut on which the subsequent computations are based is measured at Wave Probe 1 situated at a distance o f 16.25 m o f f the centieUne o f the basin, close to the track of the passing vessel. The duration o f the wave-cut measurement was restricted so that only outgoing waves would be measured. The measurement was stopped at a point when reflections from the basin side were becoming apparent.

For the extrapolation , the time record was padded with zeros to 100 s. I n Fig. 4 the measured and computed wave elevations at Wave probe 1 are shown for the first 30 s.

C e n l e r l l n e t a n h

116.3

^ 12.5 n

Figure 3 : Set-up in the Shallow Water Basin. All values are full scale

In this case, however, high frequency components obtained from the Fourier analysis of the wave cut are ignored since they do not contribute significantly to the wave forces on the barge. Leaving these components out of the inverse transform leads to small differences with the input wave cut.

0 . 4 , — • — — • — — • — . — — — • — -Fourier reconstruction Measured

Time (s)

Figure 4 : Measured wave elevation and its Fourier reconstruction

Figure 5 shows the wave directions and amplitudes of the Fourier components of the measured Wave Probe 1 record. I n this Figure, the wave direction o f 0 degrees signifies wave components with crests at right-angles to the sailing direction of the ferry (transverse waves) and 90 degrees signifies wave components with crest direction parallel to the basin axis.

Frequency {f/s)

Figure 5 : Direction and amplitude of frequency components of tlie measured wave cut

The total number o f Fourier components amounted to 652 between 0 r/s and 5.0 r/s. The bulk o f the energy in the wash waves lies between 0.5 r/s (12.5 s) and 4.0 r/s (1.6 s).

Prediction of wave elevations at other locations Based on the wave directions, amplitudes and phase angles o f the Fourier components o f Wave Probe 1, we can predict the wave elevations for other wave probes. The results o f the predictions are compared with the measured wave elevations i n Fig. 6 using 80 s data f o r the measured wave elevations i.e. until basin wall reflections are suspected. This need only be considered for Wave 1 . The computations o f the wave elevations at other locations have been carried out by means o f 3-d diffraction computations with panel models o f the barge and vertical basin side next to the captive barge model. This means that the results o f both measured and computed wave elevations at Wave Probe 2 through 6 include reflections both f r o m the basin side and the barge model. The basin side was modelled by panels over a length o f 2000 m (1000 m before and 1000 m after the location o f the array o f wave probes).

The results shown in Fig. 6 confirm that the wave elevations are well-predicted with some deviations f o r the shorter wave components. These components are relatively steep and showing some non-linear behaviour by the fact that the measured peaks arrive just ahead o f the peaks predicted by linear theory. This time shift appears to increase with the distance between wave probe land wave probe 4. Also the differences i n the f o r m o f the short wave components increase with distance. Due to the limited duration o f the wave records, no short wave components are seen in Wave 5 and Wave 6.

Noteworthy is the fact that as the distance f r o m the passing vessel increases, the wave elevation records show longer period wave components first, followed by the shorter components. This is a manifestation o f the dispersion o f the waves.

0 20 40 60 60 100 Time (s)

Figure 6 : Measured and predicted wave elevations Wave 6 in Fig. 6 shows a consistent phase shift between measured and computed wave not seen to the same degree in other wave elevations. This could be due to a slighty different position of the wave probe i n the experiments and the position used for the computations. Prediction wave forces on the captive barge

The wave forces on the captive barge have been measured for a sufficientiy long duration for the wash wave pattern to refiect back and forth between the vertical basin walls. Consequendy, i n order to be able to compare computations with the measurements, due regards must be taken o f those wave refiections .

Figure 7 shows , schematically, the wave pattem generated by the passing vessel and the location o f the captive barge. It shows diverging wave crests moving in the duection o f the vertical basin side (green arrow) and their reflections from the same side (no arrow). The wave crest indicated by W l denotes the (short) wave record recorded by Wave Probe 1 up to the time that reflection are expected at that location. See Fig. 4, Wave 1. Having modelled the vertical side o f the basin on the same side as the barge, we need to include i n Üie computational model not only the wave measured but also of subsequent reflected wave fronts which travel i n a direction towards the barge side o f the basin.

These are denoted by 3, 5, etc i n Fig. 7. Since the vessel is sailing on the basin centreline and both basin sides are vertical reflecting walls, we can obtain the record for Wave Probe 1 including those components o f its reflections which travel towards the same basin side. This is obtained by adding the relevant wave record intervals to the record o f Wave Probe 1. The relevant wave record intervals can be obtained f r o m predictions based on Wave Probe 1 assuming unrestricted width o f the basin, i.e. open water conditions. See Fig. 8.

Figure 7 : Scliematic representation of reflected wave patterns in tlie basin

In this Figure the crest-line o f W l is schematically extended. From Snellius' Law we know that the wave elevation record 3 is equivalent to elevation record 3a , the record at 5 equivalent to 5a etc.

Figure 8 : Determination of reflected wave records from the first wash wave front

These records are easily obtained based on the Fourier components of Wave Probe 1 taking into account the locations of points 3,5,7 etc. relative to the location o f Wave Probe 1.

Such a record cannot be compared direcdy with measurements since i t only contains waves going i n the direction of the barge side o f the basin while measured wave records also contain wave reflected f r o m that side. Having obtained the extended wave record in this way , diffraction computations can be carried out to deteimine the wave forces on the captive barge including the effect of the wall next to the barge model.

This has been carried out f o r different extensions o f the record o f Wave Probe 1 i.e. W l , W1-I-W3, through to W1-I-W3-I-W5+W7+W9-HWI1. Where W l , etc, denotes the wave elevation records valid at locations 1,3,5 etc. shown i n Fig. 7.

- Computed . W l

- Measured

Figure 9 : Measured and predicted wave forces using W l as input

The results f o r the forces and moments on the captive barge are shown i n Fig. 9 through Fig. 11. The measured records have a duration o f about 260 s f u l l scale while the computed records are given for 300 s. The measured forces show some spikes , mainly at the beginning o f the records. Figure 9 shows the comparison for the case that the input wave is restricted to the initial record o f Wave Probe 1, i n this case denoted by W l in the Figure. The surge and sway forces and the pitch and yaw moments compare reasonably well with measurements. The measured heave motions , while showing comparable wave frequency components , also contains l o w frequency components not present in the computed results. These are most likely due to ' draw-down' effects which do not obey the free wave dispersion relationships which underlies the present method. Draw-down effects are related to the primary flow about the vessel.

The prediction for the roll moment is considerably below the measured values. No definite explanation can be given at this time. A possible reason could be the position of the roll moment axis i n computations and experiments and not due to any significant differences i n the waves since other force modes correspond quite well, see, f o r instance , the sway force.

Regarding the time interval over which the records are comparable we see that beyond 100 s results completely fail to correlate. This is due to the fact that the measured data includes the effects o f the wave reflections i n the basin which are not present in W l (Wave Probe 1) data. When we increase the number of incoming wave components to W 1 + W 3 , shown i n Fig. 10, we see that the con-elation is generally comparable to results shown in Fig. 9 but the interval over which the correlation is valid is now increased to approximately 170 s. Finally,

for brevity, we show the results for the case of W1+W3+W5+W7+W9+W11 in Figure 11. Here again we see that the quality of the correlation is more or less the same but the interval has now increased to about 260 s. A t this time the vessel which is sailing at 34 kn has travelled about 4.5 k m past the barge. Figure 11 shows that the correlation between measurements and computations are generally of the same quality from beginning to end which is a convincing argument for the validity of the hnear wave model and i n line with Whittaker's conclusion that Unear dispersion is valid for the behaviour of wash waves o f fast vessels.

Application of the criteria for motions and accelerations.

In the first part of this paper and overview was given o f the development of new criteria f o r the peimissible wash wave induced motions of vessels moored on inland waterways. In this section we will apply the computational method for the wave-induced forces and motions to the free-floating barge discussed above. We have selected, as an example, location A on the moored barge, see Fig. 3, for which we have computed the heave motions and total accelerations (square root o f the sum of the squares of x,y and z acceleration components) using the wave results f o r the passing Fast Ferry treated i n the afore-going part.

The same 3-d diffraction computations which produced the wave forces discussed i n this paper also produced time records of the free-floating motions of the barge. The results for the heave motions and total acceleration of point A are given in Fig. 12 and Fig. 13 respectively for two cases i.e. for the case o f the wave input W l and f o r t h e c a s e o f W l + W 3 .

The results show that already with only W l as input, the peak values of the heave motions and absolute accelerations have been reached. With W1+W3 as input a shghtly higher peak acceleration value is found at about 170 s.

It is not expected that for longer durations , for instance , for the case of W 1 + W 2 + W 3 + W 5 + W 7 + W 9 + W 1 1 the motion or acceleration results would be significantly different since , due to dispersion , the longer waves, which are mainly responsible f o r the motions, would most likely show lower extremes. This may not be the case for the shortest and steepest wave components which , from observation of f u l l scale wash waves f r o m fast vessels, tend to stay together due to non-Unear effects. Such waves, however, have littie effect on the barge treated i n this paper due to its dimensions.

Figure 10 : Measured and predicted wave forces using W1+W3 as input

' ' ^ 1—I - Compuled.W1+W3+W54-W7+W9+W11 I —

Figure 11 : Measured and predicted wave forces using W1+W3+W5+W7+W9+W11 as input

Luth et al , 2009, Ref [2] give two levels of criteria for the motion and acceleration limits for inland waterway barges:

1. For manual work on a barge the motions of the barge should be restricted to:

- A vertical displacement of 0.45 m - A total acceleration of 0.6 m/s^

2. For activities related to loading and unloading (for instance hoisting by crane and boarding and unboarding a vessel) i t is recommended to restrict the motions to : - A vertical displacement of 0.30 m

0.15

0.10

Figure 12 : Computed lieave motions in point A on tlie barge

200

Figure 13 : Computed total acceleration in point A on the barge

Viewing the results given i n Fig. 12 for the heave motions and Fig. 13 f o r the total acceleration we see that in this case all criteria are met i.e. this particular passing situation will not lead to excessive motion behaviour o f the moored barge.

CONCLUSIONS

In this paper we have demonstrated the use of a method to analyse the effects of wash waves of a fast vessel sailing at 34 kn on the loads and motions of a pontoon moored i n restricted water. The method is based on using a measured or computed wave-cut taken close to the track of the passing vessel. In this paper, the input from the wave cut measured during model tests is combined with a standard frequency-domain linear 3-d diffraction code through the application of Fourier transforms to connect time- and frequency-domain

results. Output of the computations are the transient wave loads and motions of the moored barge.

Comparisons of computed and measured wave elevations at different locations have shown that linear wave theory can predict wave propagation quite accurately. Some evidence of non-linear effects in the waves are seen in the slightly different dispersion of shorter, steeper wave frequency components.

The computed wave forces, except the roll moment on the captive barge are shown to be well predicted by the method followed. The good correlation between the measured and computed wave forces and moments is maintained long after the fast vessel has passed the location of the barge confirming the validity of linear wave theory for predicting wave propagation over longer distances.

The free-floating , transient motions of the barge were computed based on the computed wave forces , added mass and damping. The motions of a corner point of the barge were subsequent computed as well as the total acceleration of the same point. The results of these computations were compared with motion criteria developed based on an extensive field investigation on passing vessel induced motions. Results show that for the particular case studied, the motions and accelerations were within recommended limits.

R E F E R E N C E S

[1] BOS, J.E., van der H O U T , I.E. and K E U N I N G , J.A.:" Golfhinder: een relatie tussen bewegingen en hinder op afgemeerde schepen", T N O report T N O _ D V 2007 C461, December 2007 [2] L U T H , H.,BOS, I.E., K E U N I N G , J.A. and van

der H O U T , 1.:" The Relationship between Motions of Moored Ships due to Wake Wash of Passing Vessels and the Hindrance thereof. International Conference on Innovation i n High Speed Marine Vessels, Fremande, Australia, 2009

[3] W H I T T A K E R , T.J.T., D O Y L E , R. and ELSA6ER, B . " A n Experimental Investigation of the Physical Characteristics of Fast Ferry Wash" 2"'' International EuroConference on High-Performance Marine Vehicles H l P E R ' O l , 2001.

[4] R A V E N , H . C ' A Solution Method f o r the Non linear Ship Wave Resistance Problem" PhD Thesis, Delft University of Technology, Delft,

1996.

[5] PINKSTER, J.A. "Hydrodynamic Interaction Effects in Waves." ISOPE'95, The Hague, 1995. [6] PINKSTER, J.A. and N A A I J E N , P.:" Predicting

the Effect of Passing Ships ", International Workshop on Water Waves and Floating Bodies , Le Croisic, France, 2003