Proceedings of

HYDRONAV 2010
Gdask, 12-13 May 2010
Organised by:

CONFERENCE ORGANIZER Centrum Techniki Okrtowej S.A.

COCO-ORGANIZERS Foundation for Safety of Navigation and Environment Protection Faculty of Ocean Engineering and Ship Technology Gdask University of Technology Department of Mechanical Engineering Wrocaw University of Technology

Centrum Techniki Okrtowej S.A., 2010 ISBN 978-83-931045-0-5

Printed by: STUDIO 4

INTERNATIONAL SCIENTIFIC COMMITTEE

Prof. L. Kobyliski - chairman Foundation for Safety of Navigation (PL) Prof. M. Atlar - University of Newcastle upon Tyne (UK) Prof. V. Bertram - Germinischer Lloyd (DE) Prof. M. Dzida - Technical University of Gdask (PL) Prof. O. Faltinsen - Norwegian University of Science and Technology (NO) Prof. A. Francescutto - University of Trieste (IT) Prof. H. Jarzyna - Institute of Fluid-Flow Machinery (PL) Prof. T. Koronowicz - Institute of Fluid-Flow Machinery (PL) Prof. J. Kulczyk - Technical University of Wrocaw (PL) Prof. Y. Nechaev - State Marine Technical University (RU) Prof. L. Perez Rojas - Technical University of Madrid (ES) Prof. F. Stern - The University of Iowa (US) Prof. J. Szantyr - Technical University of Gdask (PL) Prof. T. Szelangiewicz - Maritime University of Szczecin (PL) Dr. L. Wilczyski - Ship Design and Research Centre S.A. (PL)

ORGANIZING COMMITTEE
Mr. W. Grski - Chairman - Head of Ship Hydrodynamics Division CTO S.A. Dr. T. Bugalski - Secretary Mr. M. Reichel - Member

PREFACE

HYDRONAV 2010 is the eighteenth in a series of conferences devoted to hydrodynamics in the ship design, safety and operation. This year it is organized by Ship Design and Research Centre S.A. in Gdask. The most significant aim of HYDRONAV conferences is to exchange the knowledge and promote discussions of innovative research and industrial applications concerning naval architecture and marine engineering. It is the intention of the HYDRONAV organizers that the recent results of theoretical and experimental studies will be reported during the discussions based on groups of related papers. Papers submitted to the conference cover various aspects of hydrodynamics: o CFD and EFD in ship hydromechanics, o ship propulsion, o seakeeping, o manoeuvrability, o stability and safety at sea. We hope that results presented and discussions conducted during the conference will bring benefits to all participants and initiate new interesting research projects or scientific collaboration.

The organizers would like to thank the authors, the chairmen, session attendances and members of the international committee for their efforts and contribution to the conference.

Organizing Committee

CONTENTS

Keynote Lecture Szantyr Jan A - The Crucial Decision Points in the Propeller Design Process 7

Session MANOEUVRING Grnicz Tomasz - Influence of bow undercut angle on ship manoeuvrability in shallow water Tabaczek Tomasz - Numerical simulation of planar motion of a twin-screw inland waterway vessel in shallow water Reichel Maciej - Manoeuvring abilities of podded ships with different stern shapes 37 51 35

Session FLOW, RESISTANCE & PROPULSION 1 Wroniszewski Pawe - Verification of various methods for calculation of diffracted wave field around the ship Kozowska Anna, Sverre Steen - Ducted and Open Propeller Subjected to Intermittent Ventilation Dymarski Pawe, Kraskowski Marek - Using the vortex generators for improving the wake flow of large ships - preliminary results of CFD analyses Streckwall Heinrich - Pre- and Postswirl Devices in View of their Optimum Adjustment to the Propeller Layout Dymarski Pawe - Experimental and Numerical Modelling of Cavitation and Erosion on Rudders 121 107 93 77 63

Session FLOW, RESISTANCE & PROPULSION 2 Kulczyk Jan, Skupie Emilia - Impact of parameters of propulsion system on fuel consumed by inland waterway vessel Bugalski Tomasz, Hoffmann Pawe - Numerical simulation of the flow around ship and rotating propeller Wilczyski Leszek - Design of Wind Tunnel for Maritime Use Dymarski Czesaw, Nakielski Jacek - Mathematical model and numerical calculations of the pelagic trawl system 167 149 161 137

Session SEAKEEPING, SAFETY & STABILITY Warmowska Monika, Jankowski Jan - Problem of water flow on deck of small vessel Kobyliski Lech - New IMO Intact Stability Code Kobyliski Lech, Nowicki Jacek, Oleksy Jerzy, Jaworski Tomasz - Drift Path Prediction of Damaged Tankers Matusiak Jerzy - On the modeling of irregular waves in the non-linear ship dynamics method Laidyn Hinz Tomasz - Probability of capsizing in dead ship condition 227 239 213 183 197

The Crucial Decision Points in the Propeller Design Process

Jan Szantyr Gdansk University of Technology, Gdansk, Poland

The paper presents an overview of the typical propeller design process with special attention paid to the crucial decision points. These points include:

initial propulsive analysis and decision about the main design parameters propeller diameter and number of blades,

decision about the detailed geometrical parameters: blade outline, rake, skewback and type of the blade section profiles,

selection of the propeller design point defined by the delivered power, ship velocity and propeller (engine) rate of rotation,

assessment of the results of the self propulsion model experiments with the designed propeller and decision about the necessity of the propeller geometry modification,

assessment of the results of the cavitation model experiments with the designed propeller and decision about the necessity of the propeller geometry modification,

For each of the above decision points the detailed discussion of the motives behind each decision together with the available options is given and the logical reasoning leading to the selection of the best option are presented.

Keywords: ship hydromechanics, marine propellers, design procedures

1. INTRODUCTION Marine propeller is a very special device, which has a decisive influence on the performance of the ship. This performance may be understood in many different ways, for example as the purely technological fact of achieving the prescribed ship speed at prescribed power and prescribed prime mover rate of rotation in the prescribed navigational condition, or as the more comprehensive economic characteristic of achieving maximum profitability of ship operation over her entire life span, i.e. over thirty or more years of sailing across all the worlds seas and oceans. In the present practice the first concept of the ship performance is used to assess the quality of marine propellers in the course of the acceptance sea trials of a newly built or extensively modified ship, although the second approach is gaining importance and special procedures are being developed to put it into practical use. However, the propeller is still designed to fulfill the prescribed sea trials requirements, which are supposed to be conducted in the standardized and controlled way. The positive results of sea trials are considered to be an unquestionable proof that the propeller has been designed and manufactured correctly. On the other hand, it is obvious that this correctly designed propeller will never again operate in the conditions identical to those of the sea trials. Therefore the logical way of action is to use the second approach to determine such requirements for the acceptance sea trials, fulfilling of which would ensure the desired ship performance during her entire life span.

2. OVERVIEW OF THE PROPELLER DESIGN PROCESS The complete process of propeller design is a complicated procedure, involving both model experiments and computations (cf. Carlton 2007, Dudziak 2008, Jarzyna et al. 1996). In certain relatively less important cases (e.g. design of propellers for small and slow boats), only some selected elements of the design process are executed and model experiments are often substituted by calculations. The block diagram of the complete propeller design process for a large ship is presented in Figure 1. In the course of execution of the design process the designer continuously accumulates more and more information about the considered case and he comes across several decision points, in which important decisions must be made, significantly influencing the final result.

Figure 1 Block diagram of the propeller design process

These important decision points are located in the following moments in the design process, as shown in Figure 1: 1. after Block 4 the propeller diameter and number of blades are defined and a suitable stock propeller is selected, 2. after Block 5 the propeller design point is selected, together with the details of propeller geometry such as blade skewback, rake and type of blade section profiles, 3. after Block 9 the modification of the designed propeller blade pitch may be necessary, 4. after Block 10 the modifications of the designed propeller blade pitch, skewback, rake and blade section profiles may be necessary.

3. SELECTION OF THE MAIN DESIGN PARAMETERS 3.1 Typical input data In this paper one selected example of a container ship is used as an illustration of the problems encountered in the course of design development. This example is based on the four Technical Reports of CTO SA (2004) listed in the References. The information for the propeller designer included in the contract between the ship owner and the shipyard may contain, for example, the following text, accompanied by data collected in Table 1 and the drawing of the ship hull lines shown in Figure 2.

2500 TEU container ship, capable to maintain a service speed of 20.0 knots at 11.2 [m] draught with Continuous Service Rating (CSR) of the main engine equal to 90 per cent of the Maximum Service Rating (MCR) in wind and sea conditions corresponding to Beaufort Scale 2 and inclusive of 15 per cent sea margin above sea trials. Basing on the analysis of similar ships the CSR is expected to be in the range of 15.000 to 15.500 [kW]. The ship is to be equipped with the single screw, direct propulsion system, with a stern-located diesel engine and a short shaft line. The engine is a two-stroke, six cylinder low speed diesel with MCR=17220 [kW] at n=104 [rpm]

Table 1: The main particulars of the ship

Length between perpendiculars Length on waterline Breadth moulded Draught fore Draught aft Volume displacement Wetted surface Block coefficient Midsection coefficient Prismatic coefficient Waterline coefficient

m m m m m m3 m2 -

Design draught 190.00 193.75 32.20 11.20 11.20 42597 7740 0.6217 0.9890 0.6285 0.8073

Ballast draught 190.00 188.92 32.20 5.70 7.31 22808 5436

Figure 2 Theoretical lines of the stern part of the ship hull

After the ship model is manufactured, the resistance model measurements are conducted in the towing tank. Usually these measurements are performed for a required number of loading conditions of the ship (ballast, full load draught, scantlings draught etc.) and for the range of ship speed covering the anticipated design speed. Example of the experimentally determined resistance curve re-calculated into full scale is shown in Figure 3.

Figure 3 Ship resistance curve determined in the model basin experiments

Simultaneously, the measurements of the non-uniform velocity are performed, including one, two or all three components of the so called nominal velocity field. Example of the measured axial component distribution is shown in Figure 4. The resistance curve should be compared with data for similar ships and if the performance of the currently analyzed case is visibly inferior, the suggestions for modification of the hull geometry (especially the bow part of the hull) should be formulated. Similarly, the degree of non-uniformity of the wake velocity distribution should be assessed, using for example the criteria formulated by Odabasi and Fitzsimmons (1978). If these criteria are not fulfilled, the suggestions for modification of the stern part of the hull may be formulated. However, both these suggestions are beyond the scope of the actual propeller design and will not be discussed here in greater detail.

Figure 4 Wake distribution behind the hull determined in the model basin experiments

3.2 Determination of the propeller diameter and number of blades The propeller diameter and number of blades are mutually correlated parameters, which must be determined on the basis of the compromise between several, often conflicting criteria and requirements. The following factors must be taken into account: The propeller diameter should fulfill the requirements of the minimum propeller hull clearances, formulated by the appropriate Classification Society Rules. The required clearances usually decrease with increasing number of propeller blades. This is logically justified, because the amplitudes of the pressure pulses and unsteady shaft forces generated by the propeller decrease with the increasing number of blades. Furthermore, some of the Classification Societes include the dependence of the

recommended clearances on the shaft power, hull block coefficient and rudder thickness, which reflects the different physical phenomena involved in generation of the propeller excitation of the hull vibration. In case of the ship serving as the example in this paper the maximum possible diameter of the five-bladed propeller determined according to the Lloyds Register of Shipping Rules is 6.8 meters.

Figure 5 Propeller hull clearances according to Lloyds Register of Shipping

The propeller number of blades should be selected considering the structure of the ship hull wake and the number of cylinders of the planned main propulsion engine in such a way that the resonance between excitations generated by the propeller and the engine is avoided. As a rule, in case of a single screw ship having upper and lower wake

peaks, propeller by the engine with an even number of cylinders, the even number of propeller blades should be avoided. Both the propeller diameter and number of blades should be selected in such a way that the highest propulsive efficiency (or the lowest power consumption) is achieved in the ship design condition. This is done in the course of the so called initial propulsion analysis, based on the experimentally determined resistance curve and on the approximate, empirical computation methods, such as that of Holtrop and Mennen (1982) and on systematic propeller series data, for example Kuiper (1992). The results of such an analysis are shown in Figure 6. In this figure the dependence of propulsive efficiency D , propeller mean pitch P/D, propeller expanded blade area ratio EAR, delivered power PD and maximum amplitude of the propeller-induced unsteady pressure field Pz on the propeller diameter for two alternative numbers of blades z=4 and z=5 is shown. The analysis of the data presented in Figure 6 leads to the following conclusions: The propeller diameter of 6.75 [m] gives a reasonable compromise between propulsive efficiency and the level of propeller-generated pressure pulses At this diameter the necessary delivered power is similar for four and five bladed propellers and it is equal to about 15300 [kW] The engine power is sufficient to ensure ship speed of 20 knots on trials at 90% of Maximum Continuous Rating and assumed 99% shaft efficiency, because it gives the delivered power equal to 17220*0.9*0.99=15340 [kW] At the diameter 6.75 [m] the four bladed propeller has the pitch coefficient P/D=0.94 (constant radial pitch distribution) and the expanded blade area ratio 0.565, while the five bladed propeller has the pitch coefficient P/D=0.92 (constant radial pitch distribution) and the expanded blade area ratio of 0.665 In some cases the initial propulsion analysis may include also other parameters, such as for example propeller mass or propeller moment of inertia. In the example considered in this paper these factors were regarded as having only secondary importance. Finally, the diameter of 6.75 [m] and number of blades z=5 were selected. Information about the mean propeller pitch and expanded blade area ratio enabled selection of the so called stock propeller, which is an existing propeller model with known hydrodynamic characteristics.

Figure 6 Summary of the initial propulsion analysis

Stock propeller in fact serves as a measuring device in the course of self propulsion model experiments, in which the more accurate values of the propulsive coefficients (mean effective wake fraction, thrust deduction and relative rotative efficiency) are obtained and the more accurate powering prediction for the ship may be computed. Example of such a powering prediction is shown in Figure 7. From this figure the predicted service speed at delivered

power of 15340 [kW] may be determined as 19.76 knots (marginally less than expected) at 99.9 rpm of the engine, while the expected ship speed on trials at the same power is predicted as 20.55 knots at 104.1 rpm of the engine. The values of the propulsive coefficients determined in the course of self-propulsion model experiments with the stock propeller serve as important input data in the final propeller design computations.

Figure 7 Powering prediction based on self-propulsion model experiments with the stock propeller

4 SELECTION OF THE PROPELLER DESIGN POINT The marine propeller may be regarded as a device linking together the main propulsion engine of the ship with the ship hull into one integrated and harmonized dynamic system. Both the operational characteristics of the engine and the resistance/propulsion characteristics of the ship hull influence the propeller design. From the engine point of view the propeller should be designed in such a way that it ensures the required optimum loading condition (power and rate of rotation) of the main engine throughout the highest possible proportion of the total ship service time. It is obvious that this optimum engine loading condition leads to the lowest specific fuel consumption. This effect may be achieved through an appropriate selection of the propeller design point. The reasoning behind the procedure for selection of the propeller design point should take into account different specific requirements of the ship owner and of the engine manufacturer. Consequently, it may look differently in different cases. The case discussed below and illustrated in Figure 8 refers to the typical low speed marine piston engine.

Figure 8 Selection of the propeller design point

Situation shown in Figure 8 refers to a different case than the example considered in the paper, but it explains the general logic. In this figure two curves showing power consumption by the propeller in service and trial condition are shown, together with the operational (service) margin and sea margin. The propeller design point PDP is located at the contractual service speed, in the middle between the trials and service power consumption lines. Such determination of the propeller design point ensures the optimum utilization of the engine power in the varying ship loading and navigational conditions

5 SELECTION OF THE DETAILED GEOMETRICAL PARAMETERS 5.1 Propeller blade outline The propeller blade area resulting from the selected blade outline influences the amplitudes of the unsteady bearing forces and the pressure pulsations generated by the cavitating propeller operating in the non-uniform inflow velocity field behind the ship hull. Increase the blade area, especially in the region covered by sheet cavitation (cf. Figure 9: A the initial blade outline, B the blade with increased area) reduces the amplitudes of the unsteady bearing forces and the pressure pulsations generated by the cavitating propeller operating in the nonuniform inflow. Elongation of the blade section chords allows also some reduction of the maximum thickness of the blade sections while retaining the required blade structural strength.

Figure 9 Influence of the blade outline on the unsteady cavitation formation

It should be also kept in mind that when the blade section chords are elongated and the (dimensional) mean line camber remains unchanged, then the camber to chord ratio is effectively reduced. Such a modification reduces the underpressure on the suction side of the blade, thus reducing the resulting maximum volume of the unsteady sheet cavity in the course of growth-decline cycle. Simultaneously, the process of cavity growth and decline is extended in time, further reducing the rate of cavity volume variation with the blade rotation. The rate of variation of the blade section lift force is also reduced. Consequently, both the pressure pulsations and the unsteady shaft forces amplitudes are reduced. Increase in the blade area may be also effective in elimination of bubble cavitation in the mid-chord region of the blade sections. The unfavorable side effect is an increase in the blade section drag due to larger area, leading to some reduction of the propeller efficiency.

5.2 Propeller radial pitch distribution Propeller radial pitch distribution is an important factor, which may be used to reduce the amplitudes of the unsteady bearing forces and the pressure pulsations generated by the cavitating. The general idea is to reduce the pitch in the tip region covered by the sheet cavity and to compensate for the resulting loss of the total blade lift by an appropriate increase of pitch in the central part of the blade (cf. Figure 10: A initial radial pitch distribution, B modified pitch distribution). The compensating increase in the local blade pitch should be carefully distributed along the blade radius in such a way, that it does not induce cavitation in the regions of the blade which were previously cavitation-free.

Figure 10 Different radial pitch distributions

Such a modification leads to reduction of the maximum volume of the unsteady sheet cavity, thus reducing the rate of variation of this volume with the blade rotation. As a consequence, the pressure pulses amplitudes induced by cavitation are reduced. Simultaneously, the maximum local lift coefficients in the outer part of the blade are also reduced, leading to some reduction of the amplitudes of the unsteady bearing forces. The limiting factor for the possible blade pitch reduction is the danger of face (pressure side) cavitation, which may appear first in the blade positions outside the upper wake peak. This form of cavitation is highly erosive and it should be avoided. Another unfavorable side effect of the pitch reduction is some loss of propeller efficiency. This loss of efficiency is rather smaller than generally expected and it should be regarded as a reasonable price for the effective reduction of the unsteady bearing forces and pressure pulsations.

5.3 Propeller blade skewback The main purpose of applying propeller blade skewback is to reduce the amplitudes of the fluctuating hydrodynamic bearing forces generated by the propeller, to minimize the unsteady cavitation phenomena on the propeller blades and to reduce the amplitudes of pressure pulses generated by the cavitating propeller. The mechanism of this effect is quite simple and it is explained in Figure 11.

Figure 11 Different radial distributions of the blade skewback

In this figure the symmetrical and skewed blade outlines are drawn against the map of the typical distribution of the axial velocity coefficient behind the single screw ship. As the propeller with symmetrical blades rotates, its all blade section enter the upper region of the highly retarded flow at the same time, leading to a sharp increase of the hydrodynamic forces on the entire blade. Furthermore, there is a corresponding sharp increase in the volume of unsteady sheet cavity on the blade, giving rise to high amplitudes of the pulsating pressure. For the skewed blade, these processes are much more extended in time, because the inner sections of the blade enter the region of the highly retarded flow first and they are already leaving it when the outer blade sections start to enter this region. Consequently, both the unsteady hydrodynamic forces and the cavity volume on the blade grow and decline much more gently than in the case of the symmetrical blade. Despite the simplicity of the above described skewback effect on the unsteady propeller performance, skewing of the blade should not be applied blindly. The right amount of skewback should be derived from a careful analysis of the interaction between the non-uniform velocity field and the propeller geometry.

5.4 Propeller blade rake Moderate raking of the propeller blades does not affect the hydrodynamic characteristics of the propeller. Rake is applied to accommodate the propeller properly at the ship stern and to ensure the recommended clearances between the propeller blades and the hull cf. Figure 12 where un-raked blade A is compared with the raked blade B. Keeping these clearances is important for controlling the pressure pulses generated by the propeller and transmitted through the water onto the ship hull. The amplitudes of these pulses are reduced roughly with the square of the distance between the tip region of the blades and the hull surface. Additionally, increasing the distance between the stern frame and the propeller reduces the non-uniformity of the inflow velocity field, thus making the operating condition of the propeller more favorable. On the other hand, an increase in rake leads to higher blade stresses, resulting from the additional bending moments of the centrifugal forces. As the consequence, the raked propeller blades are slightly thicker than the corresponding un-raked propeller blades. This may lead to a minimal increase in the hydrodynamic drag and a slightly reduced propeller efficiency. In case of highly skewed propellers often a rake-forward is applied in order to reduce the excessive length of the propeller resulting from the high skew.

Figure 12 Different radial blade rake distributions

5.5 Blade section profiles In the contemporary propeller design the so called combined blade section profiles are frequently used. These sections result from the superposition of mean lines and chord-wise thickness distributions. Within the limits of the linear theory, the hydrodynamic properties of both these components may be also superimposed. In this sense the mean line is responsible for the lift, efficiency and cavitation properties of the blade section, while the thickness distribution is responsible for drag, structural strength of the blade section and also for its cavitation properties, especially in the leading edge region. Reasonable selection of both components of the blade section requires an access to the results of systematic wind tunnel tests with profiles, for example in Abbott and Doenhoff (1959). The effect of the chord-wise thickness distribution, especially near the leading edge, is also very important for the cavitation properties of the blade sections. In order to obtain a complete picture this effect must be superimposed over the effect of the mean line. The analytically designed chord-wise symmetrical thickness distribution, operating at zero angle of attack, generates the chord-wise pressure distribution without a peak at the leading edge (profile A in Figure 13).

Figure 13 Effect of the leading edge radius on the pressure distribution on a profile

Of course when this thickness distribution is placed at a non-zero angle of attack, the underpressure peak appears at the leading edge. If the leading edge radius and the thickness distribution close to the leading edge are increased (profile B in Figure 13), the underpressure peak is present even at the zero angle of attack. However, the increase of this underpressure peak with the growing angle of attack is now much smaller than in the case of profile A. The conclusion is obvious: the blade section profiles for operation in axi-symmetrical velocity field (i.e. at constant angle of attack), should have possibly sharp leading edges, while for operation in circumferentially non-uniform inflow velocity field a certain increase in the leading edge radius of the blade section profiles may be beneficial.

5.6 Design of the propeller The actual design calculation of the propeller is performed by an appropriate computer program (cf. for example Szantyr 1987), taking into account all accumulated information and all decisions already made (propeller design point, blade outline, blade pitch distribution, rake, skewback and type of blade section profiles). Contemporary propeller design programs usually operate in an interactive mode, enabling easy modifications of the detailed propeller

geometry in the course of calculations. As the result the detailed geometry of the designed propeller is obtained (see Figure 14 and Table 2) and the model of the designed propeller is manufactured. This model is now used for self-propulsion model experiments in the towing tank and for cavitation experiments in the cavitation tunnel

Figure 14 Drawing of the model of the designed propeller

Table 2. Radial distribution of the geometrical parameters of the designed propeller

r/R 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 Where:

[deg ] c[mm] P/D 0.7403 49.67 1254.3 0.8317 41.43 1494.3 0.9022 35.68 1683.9 0.9559 31.32 1822.6 0.9830 27.54 1904.6 0.9735 23.88 1918.9. 0.9315 20.34 1859.7 0.8795 17.28 1643.0 0.8362 14.90 320.0 r/R non-dimensional radius
P/D pitch coefficient pitch angle [deg] c section chord length [mm] S skewback ordinate [mm]

S [mm]
-20.8 -131.0 -242.2 -296.4 -257.5 -115.3 119.8 442.8 895.0

Z R [mm]

t [mm]

m[mm]

rLE [mm]

0.0 323.9 35.4 16.2 0.0 262.8 52.5 13.1 0.0 213.8 61.0 10.7 0.0 174.1 61.4 8.7 0.0 141.1 55.4 7.1 0.0 112.3 45.5 5.6 0.0 85.0 33.2 4.4 0.0 56.7 18.4 3.0 0.0 24.6 0.3 3.0 t maximum section thickness [mm] m maximum section camber [mm]

rLE - leading edge radius [mm]

Z R - rake ordinate [mm]

6. ANALYSIS OF THE SELF-PROPULSION MODEL EXPERIMENTS

Figure 15 Propulsion prediction based on self-propulsion model experiments with the designed propeller

The analysis of the powering prediction with the designed propeller shows that the ship speed predicted for the design condition is identical with the value obtained in the self-propulsion tests with the stock propeller 20.50 knots. Similarly, the ship speed predicted for the service condition is almost identical with that obtained with the stock propeller 19.70 knots. However, the propeller rate of rotation predicted for the trials condition is too low 103.5 rpm versus required 104.0 rpm. This shows that the propeller is marginally hydrodynamically too heavy, i.e. it generates too high torque on the engine shaft. In order to correct this situation a reduction in propeller pitch of 0.6 per cent was decided, as shown in Table 3.

Table 3. Originally designed and corrected pitch of the designed propeller

r/R 0.200 0.250 0.300 0.400 0.500 0.600 0.700 0.800 0.900 0.950 0.975 1.000

P/D original 0.7403 0.7896 0.8317 0.9022 0.9559 0.9830 0.9735 0.9315 0.8795 0.8571 0.8468 0.8362

P/D corrected 0.7356 0.7846 0.8264 0.8964 0.9498 0.9766 0.9673 0.9256 0.8739 0.8516 0.8413 0.8308

Such a small correction does not justify manufacturing of the new, corrected propeller model for the cavitation model experiments. The effect of modification may be simulated by an appropriately defined operating condition in the cavitation tunnel. Moreover, the values of the propulsive coefficients obtained in the tests with the designed propeller should be compared with those determined in the tests with stock propeller and used in the propeller design calculations. If the differences between them are significant, the design calculation must be repeated and the new propeller model must be manufactured if necessary.

7. EXPERIMENTAL AND COMPUTATIONAL ANALYSIS OF THE PROPELLER OPERATION IN THE NON-UNIFORM INFLOW FIELD 7.1 Observations of cavitation The purpose of model experiments in the cavitation tunnel is to observe and register cavitation phenomena on the designed propeller in the prescribed operating conditions and to measure the pressure pulsations generated by the cavitating propeller in the selected points on the stern part of the ship hull. In order to reproduce the propeller operating conditions behind the ship hull, the three-dimensional velocity field measured at propeller location in the model basin must be now generated in the cavitation tunnel. In the medium size cavitation tunnel, as the one employed in the presented case, a deformed ship model hull, called the dummy body, is placed in the test section of the cavitation tunnel. Example of the observed and registered cavitation phenomena is shown in Figure 16.

Figure 16 Cavitation phenomena on the propeller model at 103.5 rpm

7.2 Measurements of propeller-induced pressure pulsations The measurements of the propeller-induced pressure pulses were performed in five points located in the aft part of the dummy body placed in the cavitation tunnel. The sketch showing the position of these points is presented in Figure 17. It is typical to locate the pressure transducers directly above the propeller (point no. 1), in front and behind the propeller in the ship plane of symmetry (points nos. 4 and 5), and both to port and starboard in the plane of the propeller (points nos. 2 and 3). In some cases a large number of transducers is used (2050), in order to measure the pressure pulsations over an extensive area of the stern hull plating. This is done when the calculation of the resultant unsteady force acting on the stern part of the ship hull is to be performed. The highest amplitudes of the propeller generated pressure pulses on the hull may usually be expected directly above the propeller or at a small distance in front of the propeller, but in some cases (including the presented example) it is located elsewhere. The actual location of this maximum depends on the history of sheet cavity development and decline.

Figure 17 Locations of points for measurement of the propeller-induced pressure pulses

The propeller generated pressure pulses measured in the course of the experiments in the cavitation tunnel are most frequently presented in the form of harmonic amplitudes for the frequencies corresponding to multiples of the blade passage frequency nz. The first three harmonic amplitudes, measured in model scale (index m in Table 4), are re-calculated into full scale (index s in Table 4).
Table 4. Model and full scale harmonic amplitudes of the pressure pulses

No. 1 2 3 4 5

A1m Pa 162 169 315 176 145

A2m Pa 11 12 8 6 12

A3m Pa 0 0 0 0 8

A1s kPa 1.131 1.180 2.193 1.228 1.012

A2s kPa 0.077 0.084 0.056 0.042 0.084

A3s kPa 0 0 0 0 0.056

7.3 Cavitation erosion tests In the erosion tests the propeller models are covered with the special soft paint and then they run in the cavitation tunnel for a prescribed time in the prescribed operating condition. Usually it is considered sufficient to run the propeller model for about 40 minutes. The generally accepted upper time limit is two hours. If after that time there are no traces of cavitation erosion, the propeller may be regarded as safe in this aspect. The propeller designed in this paper showed no erosion, but for the sake of illustration, the picture of another propeller model exhibiting traces of erosion in the blade tip area is shown in Figure 19.

Figure 19 Results of the erosion paint tests with a propeller model

7.4 Computations of the propeller performance in the non-uniform inflow The computational analysis of the propeller operation in the non-uniform velocity field is performed by means of the computer program based on the unsteady deformable lifting surface theory (Szantyr, 1994). The purpose of this analysis is to determine the extent of sheet, bubble and vortex cavitation on the blades in their different angular positions and to calculate the unsteady hydrodynamic forces on the blades and on the propeller shaft, together with the propeller-induced pressure pulses. The input data for the analysis consisted of the propeller geometry (cf. Section 4), the non-uniform velocity field determined in Section 3 and the propeller operating condition as determined in the propulsion prediction of Section 6. The selected results of computations are presented in Figures 20, 21 and 22. As can be seen in these figures the computational method is capable of predicting sheet, vortex, bubble and intermittent cavitation in a quite realistic way. Moreover, it delivers the values of timedependent blade and shaft forces, which were not measured in the cavitation tunnel experiments.

Figure 20 Calculated cavitation phenomena on the propeller blade

Figure 21 Calculated hydrodynamic force and moment components on the single blade

Figure 22 Calculated harmonic amplitudes of the unsteady hydrodynamic force and moment components on the propeller shaft

7.5 Analysis of the propeller performance in the non-uniform velocity field The important assessment of the hydrodynamic quality of the hull-propeller-rudder configuration is performed on the basis of the propeller performance in the non-uniform velocity field behind the ship hull (including the rudder effect if considered important). A number of different criteria for this assessment have been developed and published for the purpose of this assessment (cf. for example Dudziak, 2008). Three aspects of the propeller operation are now analysed: forms and extent of the cavitation phenomena, pulsations of pressure generated by the cavitating propeller on the hull surface in the stern region, amplitudes of the fluctuating shaft forces generated by the propeller.

The assessment may be performed on the basis of the results of model experiments in the cavitation tunnel, on the basis of the appropriate calculations, or the combination of both.

Forms and extent of the cavitation phenomena There are no established, precise criteria for determination of the acceptable intensity of cavitation on the typical cargo ship propeller. However, there are some general rules which should be adhered to: face (pressure side) sheet cavitation is not acceptable, as it is highly erosive, cloud cavitation should be avoided, because it leads to erosion, bubble cavitation in the midchord region of the blade sections should be avoided, because it also leads to erosion; especially erosive is bubble cavitation in the blade root area, tip vortex cavitation should be avoided, because it generates intensive hydro-acoustic signals, moreover it often leads to erosion of the rudder; however, a short tip vortex being a natural continuation of the sheet cavity is acceptable, small to moderate extent of unsteady sheet cavitation on the back (suction side) of the blade is acceptable, as long as it does not generate excessive pressure pulsations and it does not affect propeller thrust, torque and efficiency; practically it means that the maximum area covered by the sheet cavity should not exceed 5 per cent of the blade area, higher frequency natural oscillations of the sheet cavity should be avoided, as they often lead to separation of the cloud cavities, which in turn generate erosion. Pulsations of pressure generated by the cavitating propeller The most popular approximate criterion for assessment of the pulsating pressure amplitudes refers to the peak-to-peak amplitude (or, alternatively, to the amplitude of the blade frequency harmonic) and it provides the following classification: very low pulsations of pressure less than 2.0 [kPa] low pulsations of pressure between 2.0 and 4.0 [kPa] moderate pulsations of pressure between 4.0 and 8.0 [kPa] unconditionally dangerous pulsations of pressure above 8.0 [kPa].

Very low values are recommended for cruise ship and ferries, as well as for some categories of research vessels. Low values are recommended for most cargo ships. It should be kept in mind that the sole acceptable value of the pressure excitation amplitude is not sufficient to ensure the vibration-free operation of the ship. The response of the ship structure to the excitation should be also considered. It is possible that a high excitation amplitude acting on the node of the specific form of vibration does not incite resonance of the structure, while a relatively small excitation acting between the nodes leads to vibration.

Amplitudes of the fluctuating shaft forces When the amplitudes of the time-dependent propeller thrust and torque are nondimensionalized by the respective mean values of these parameters, the following simple criteria may be used: very low fluctuations of thrust and torque below 0.02, low fluctuations of thrust and torque between 0.02 and 0.04, moderate fluctuations of thrust and torque between 0.04 and 0.08 acceptable on the condition that their frequency is sufficiently distant from the resonance frequency of the propeller-shaft-engine system, unacceptable fluctuations of thrust and torque above 0.08.

In the case serving here as an example all the cavitation phenomena on the propeller blades were very limited and the values of the unsteady shaft forces and the propeller-induced pressure pulses on the hull were well within the acceptable limits. Consequently, no modifications of the designed propeller geometry were necessary at this stage. If such modifications were necessary, the possibility of modifications of the propeller blade outline, pitch distribution, skewback, rake and details of the blade section profiles, as explained in 5, should be employed in order to generate appropriate changes in the propeller performance.

8. CONCLUSIONS The paper presents an overview of the typical propeller design process for a single screw cargo ship. This process contains important decision points, in which the designer must take correct decision in order to arrive at the optimum final design. These decisions are based on the accumulated experience of the designer, but first of all on the continuously collected information about different aspects of the propeller design under development. This information is acquired through a combination of computations and model experiments. For each of the four major decision points the criteria and motivation of the respective decisions were described and discussed in detail. This discussion should enable the propeller designers to take right decisions and to develop better propeller designs in the future.

ACKNOWLEDGEMENTS The author would like to express his sincere gratitude to the Management of the Ship Design and Research Centre CTO SA for their kind permission to use the selected research reports concerning propeller design in preparation of this paper.

REFERENCES Abbott I.H., Doenhoff A.E.(1959): Theory of Wing Sections, Dover Publ. Inc. New York Carlton J.,(2007), Marine Propellers and Propulsion, Elsevier Publishers Dudziak J.,(2008), Ship Theory (in Polish), Foundation for Promotion of Shipbuilding Industry Gdansk Holtrop J., Mennen G.G.J,(1982),: An Approximate Power Prediction Method, International Shipbuilding Progress Vol. 29, No. 335 Jarzyna H., Koronowicz T., Szantyr J.A.,(1996), Design of Marine Propellers Selected Problems, Ossolineum Kuiper G., (1993), The Wageningen Propeller Series, MARIN Publication 92-001 Odabasi Y.A., Fitzsimmons P.A., (1978), Alternative Methods for Wake Quality Assessment, International Shipbuilding Progress Vol. 25, No. 282 Szantyr J.A., (1987), An Interactive Program for Design of Ship Propellers, The Naval Architect, No.10/87 Szantyr J.A, (1994), Method for Analysis of Cavitating Marine Propellers in Non-uniform Flow, International Shipbuilding Progress, Vol. 41, No. 427 Technical Report CTO, (2004), Results of Model Tests at Design and Scantling Draught with Stock Propeller, No. RH-2004/T-017E Technical Report CTO, (2004), Design of New Propeller, No. RH-2004/T-018E Technical Report CTO, (2004), Results of Model Tests at Design and Scantling Draught with Designed Propeller, No. RH-2004/T-031E Technical Report CTO, (2004), Cavitation Tests and Pressure Measurements, No. RH2004/T-035E Technical Report CTO, (2004), Speed Trials and Power Measurements, No. RK-2004/B-084E

CONTACT DETAILS Jan Szantyr, Gdansk University of Technology, Narutowicza 11/12, 80-233 Gdansk, Poland jas@pg.gda.pl

Influence of bow undercut angle on ship manoeuvrability in shallow water

Tomasz Grnicz, Wrocaw University of Technology

The recent development of computer aided ship design has enabled improvements in ship manoeuvrability properties during the early phase of the design project. A new modern method for predicting ship manoeuvrability must be designed, because until recently, ship manoeuvrability was typically considered to be of secondary importance. This paper builds on an earlier paper where the author described a novel method for calculating the hull hydrodynamic coefficients. This paper describes the modelling of the method using a computer program to predict the resulting ship manoeuvrability. A series of CFD calculations for a simplified standard inland hull shape were made. The calculations were performed for different bow undercut angles and three water depths. The results from the simulation of standard IMO manoeuvres and an analysis of ship behaviour in shallow water is presented.

Numerical simulation of planar motion of a twin-screw inland waterway vessel in shallow water
Tomasz Tabaczek Wrocaw University of Technology, Wrocaw, Poland

Equations of motion and mathematical models of external forces were applied in numerical simulation of manoeuvring motion of inland waterway vessel in shallow water. Coefficients in mathematical models were determined based on data from captive model tests. Results of simulations are compared to data from free-sailing model tests of considered twin-screw twinrudder vessel.

Keywords: twin-screw vessel, manoeuvrability, numerical simulation

1. INTRODUCTION Since many years, and especially since 2002 when International Maritime Organization has adopted the Resolution MSC.137(76) Standards for Ship Manoeuvrability, a lot of research work was devoted to accurate simulation of ship motion. The goal of the present author was to predict the motion of twin-screw inland waterway vessel during manoeuvres in shallow water. The author chose an available mathematical model of ship motion in 3 degrees of freedom. Captive model tests with application of planar motion mechanism (PMM), as well as free-sailing model tests were designed and carried out specially for the purpose of present research. Coefficients for mathematical model of ship motion were determined using the data from PMM tests. Finally, the mathematical model was applied to simulate the evasive action manoeuvre. Approach and sample results of research are presented in this paper.

2. MATHEMATICAL MODEL The mathematical model developed at Kyushu University (Kijima et al., 1990), known as Kijimas model, is one of the best described in open literature. It is based on 3 DOF equations of ship motion. In coordinate system with origin located in ships centre of gravity (Fig.1) the equations take the following form:
m (udot - vr) = X m (vdot + ur) = Y Izz rdot = N (1)

where: u, v x- and y -axis components of ship speed, r yaw rate, X, Y x- and y -axis components of external force acting on ship (longitudinal and side forces), N external yaw moment acting on ship, m mass of ship, Izz moment of inertia of ship, udot = du/dt longitudinal acceleration of ship, vdot = dv/dt transverse acceleration of ship, rdot = dr/dt rotational acceleration of ship.

y0 U

G
r, N

x0

Figure 1. Co-ordinate system

Modular model of external forces X, Y and N is based on the assumptions of Mathematical Modelling Group (MMG) (Bulletin of the SNAJ, 1985). Because the model was originally developed for a single screw vessel, it was extended to include additional effects of twinscrew twin-rudder arrangement:
m (udot - vr) = XH + XP + XR m (vdot + ur) = YH + YR Izz rdot = NH + NP + NR (2)

XH, YH, NH denote hydrodynamic forces acting on ship hull during manoeuvring motions, XP, NP - forces exerted on ship hull by propellers, and XR, YR, NR - forces exerted on ship hull by rudders.

2.1 Hull hydrodynamic Forces Following Kijima et al. (1990), the cubic model of hydrodynamic forces was applied in present research:
XH = -mxudot + myvr + LdU2Xrrsin - RT(u) YH = -myvdot mxur + LdU2 [Y+Yrr+Y||+Yrrr|r|+(Yr+Yrrr)r] NH = -Jzzrdot + L dU [N+Nrr+N||+Nrrr|r|+(Nr+Nrrr)r]
2 2

(3)

where: mx, my x- and y -axis components of added mass of ship, Jzz added moment of inertia of ship, density of water, L ship length, d ship draught,

U ship speed (U = (u2 + v2)1/2), r non-dimensional yaw rate (r=rL/O), drift angle, RT(u) ship resistance, Xr , Y , Y , Yr , Yrr , Yr , Yrr , ... , Nrr

hydrodynamic coefficients of bare hull.

2.2 Propeller forces Because of asymmetry of propeller inflow between starboard and port side propellers during ship manoeuvres, each propeller is considered separately:
XP = XPS + XPP NP = -yP XPS + yP XPP

where: XPS longitudinal force exerted on ship hull by starboard propeller, XPP longitudinal force exerted on ship hull by port side propeller, yP transverse distance from ships plane of symmetry to propeller axis. In comparison to a single-screw arrangement the present model accounts for additional moment due to difference in loading of propellers. Side force due to oblique inflow is neglected similarly as in the case of single-screw ship. Thrust of starboard propeller is approximated with the following formulae:
XPS = (1 tP) TS TS = KT n2D4 KT = b0 + b1J + b2J2 J = U cos (1 - wPS) / nD = u(1 - wPS) / nD

where: tP thrust deduction factor, TS propeller thrust, KT thrust coefficient (calculated from open water characteristics approximated with square polynomial), n propeller revolutions, D propeller diameter, wPS effective wake fraction in propeller disk. Data from model tests revealed the significant variation of wake fraction wP with drift angle and even with rudder angle .

2.3 Rudder forces Similarly as in above model of propeller force both rudders are considered separately also in the following model of rudder force:
XR = XRS + XRP YR = YRS + YRP NR = NRS + NRP

Forces exerted on ship hull by starboard rudder are calculated according to the following formulae:
XRS = -(1 tR) FNSsinS YRS = -(1 + aH) FNScosS NRS = -(xR + aHxH) FNScosS - yRXRS

where: FN normal component of rudder force (positive to port side), rudder angle (positive to starboard), tR coefficient that accounts for additional drag of hull due to rudder deflection, aH ratio of additional lateral force, xH longitudinal co-ordinate of the centre of additional lateral force, xR longitudinal co-ordinate of rudder force, yR transverse distance of rudder axis from ships plane of symmetry.
FN = AR UR2 cN

AR UR cN

rudder area, effective rudder inflow velocity, normal force coefficient.

cN = [6,13/(+2,25)] sinR R = - R R = 2xRr

R hR

effective rudder inflow angle, aspect ratio of rudder ( = AR/hR2), rudder height.
UR2/U2 = (1- wR)2 [1 + Cg(s)] g(s) = KE[2 (2-KE)s]s / (1 s)2 = D/hR KE = 0,6(1 wP)/(1 wR) s = 1 - (1 wP)Ucos/nP = 1 - J/(P/D)

wR

effective wake fraction at rudder location,

C s P

coefficient that accounts for asymmetry in propeller slipstream, propeller slip ratio, propeller pitch, flow straightening coefficient.

3. MODEL SHIP AND PMM TESTS Hull form of model ship chosen for tests and simulations is presented in Fig.2 and main particulars are given in Table 1. All tests and simulations were carried out in shallow water at h/d=1,89.

WL

BL
Figure 2. Hull form of model ship and arrangement of propellers and rudders

Table 1. Main particulars of considered vessel

LPP B d D

Ship 85.50 m 11.45 m 2.65 m 2213 m3 1.745 m

Model 3.920 m 0.525 m 0.122 m 0.213 m3 0.080 m

All model tests were carried out in Ship Design and Research Centre in Gdask. PMM tests were preceded by resistance test and simple straight ahead free-sailing test in order to adjust propeller revolutions at given model speed. Because of small diameter of model propellers the open water characteristics of propeller was approximated with characteristics of a bigger geosim (D=0.148m). Subsequent captive model tests comprised oblique towing, pure sway, pure yaw and yaw with drift tests of bare hull, and oblique towing with rudders and running propellers. During tests there were measured all magnitudes (position, velocities, forces) necessary to determine coefficients in mathematical model.

Finally, free-sailing model tests of evasive action manoeuvre were carried out in towing tank. During tests the yaw rate of model ship was measured for the purpose of control of rudder executions. Propeller thrust and rudder force were measured for validation of numerical simulations.

4. COEFFICIENTS IN MATHEMATICAL MODEL OF EXTERNAL FORCES 4.1 Coefficients in mathematical model of hull forces The coefficients in model of hull forces, usually called hydrodynamic coefficients, were determined using data from captive model tests. Equations of motion (2) in conditions of captive test with bare hull take the form:
m (udot - vr) = XH + XC m (vdot + ur) = YH + YC Izz rdot = NH + NC

where XC, YC, NC denote external forces exerted by PMM guides on model ship, measured during tests. After substitution of hull forces XH, YH and NH with mathematical models of hull forces (3):
m(udot-vr) = -mxudot + myvr + LdU2Xrrsin - RT(u) + XC m(vdot+ur) = -myvdot - mxur + + LdU2[Y+Yrr+Y||+Yrrr|r|+(Yr+Yrrr)r] + YC Izzrdot = -Jzzrdot + + L2dU2[N+Nrr+N||+Nrrr|r|+(Nr+Nrrr)r] + NC (4)

During oblique towing u=const., v=const. and r=0, and equations become reduced:
0 = LdU2 [Y + Y||] + YC 0 = L2dU2 [N + N||] + NC

Four coefficients: Y, Y, N, N were determined using least square method to fit above equations to test data. During pure sway tests u=const. and r=0. Equations (4) become reduced to:
mvdot = -myvdot + LdU2 [Y + Y||] + YC 0 = L2dU2 [N + N||] + NC

Only added mass my was determined from pure sway tests.

During pure yaw tests v=0, =0, vdot=0, and:

m ur = -mxur + LdU2 [Yrr + Yrrr|r|] + YC Izz rdot = -Jzz rdot + L2dU2 [Nrr + Nrrr|r|] + NC

From pure yaw tests the following coefficients were determined: Yr , Yrr , Jzz , Nr and Nrr. In the case of yaw with drift tests no term in equations (4) can be abandoned. All crosscoupling coefficients: Xr , Yr , Yrr , Nr , Nrr were determined using the before determined values of remaining coefficients.

Table 2. Values of coefficients used in subsequent simulations

Coeff. mx my Jzz Xr

Value 14.5 kg 342 kg 171 kgm2 0.1475

Coeff. Y Yr Y Yrr Yr Yrr

Value 0.08936 0.03516 2.8389 0.01277 -0.3771 0.6961

Coeff. N Nr N Nrr Nr Nrr

Value 0.1896 -0.03744 0.1909 -0.02173 -0.8983 0.04076

The ability of mathematical model to reproduce test data is illustrated in Fig.3.

30 20 10 XC [N] 0 0 -10 -20 -30 t [s] 5 10 15 test approx.

YC [N]

40 20 0 0 -20 -40 -60 -80 t [s] 5 10 15 test approx.

60 40 20
NC [Nm]

0 -20 -40 -60 -80 -100 t [s] 0 5 10 15 NC approx.

Figure 3. Data from PMM tests approximated using mathematical model of hull forces (yaw with drift, =10deg)

4.2 Coefficients in mathematical model of propeller forces Wake fraction in propeller disk wP was determined based on measurements of propeller thrust during oblique towing, open water propeller characteristics and assumption of thrust identity. The determined values of wake fraction reveal the significant variation of wake fraction with drift angle and even with rudder angle. The arrangements of hull, propellers and rudders in extreme test conditions are shown in Fig.4. In simulations the wake fraction wPS was approximated as shown in Fig.5. Drift angle was substituted with drift angle at propeller location P:
P = xPr

where xP is a non-dimensional longitudinal co-ordinate of propeller disk (xP=xP/L).

Figure 4. Extreme positions of hull, propeller and rudder during oblique towing with running propeller ( = 16deg, = 45deg)

= 8 deg 16 deg 0 deg

0,3 0,25 0,2 0,15

wP

-8 deg

0,1 0,05

-16 deg -60 -40 -20

0 0 -0,05 -0,1 [rad] 20 40 60

Figure 5. Approximation of wake fraction in propeller disk of starboard propeller

4.3 Coefficients in mathematical model of rudder forces Coefficients that account for the effect of rudder on hull (tR, aH, xH) and for the effect of hull on rudder ( and wR) were also determined using data from oblique towing with running propellers. Equations of ship motion written for the conditions of oblique towing takes the following form:
XH + XP + XR + XC = 0 YH + YR + YC = 0 NH + NP + NR + NC = 0

According to mathematical model of rudder forces with S = P = :

XR = -(1 tR) (FNS + FNP)sin YR = -(1 + aH) (FNS + FNP)cos NR = -(xR + aHxH) (FNS + FNP)cos - yR(1 tR)(-FNS + FNP)sin

Normal forces on rudders (FNS and FNP) were measured during oblique towing. After substitution, the equations of motion can be written in the following form:
(1 tR) (FNS + FNP)sin = XH + XP + XC (1 + aH) (FNS + FNP)cos = YH + YC (xR + aHxH) (FNS + FNP)cos + yR(1 tR)(-FNS + FNP)sin = NH + NP + NC

Hull and propeller forces were evaluated using the mathematical models described in section 2. Thrust of starboard and port side propellers (TS and TP) as well as PMM forces (XC, YC, NC) were measured during oblique towing with running propeller. The following values of coefficients were determined: 1-tR = 1.18 , aH = 0.44 , xH = -0.41 . Flow straightening coefficient was determined using rudder normal force FN measured during oblique towing with running propellers. Forces measured at different drift angles were used to determine rudder angles where FNS=0 and R=0. From the mathematical model:
R = (FN=0) - R = 0

During oblique towing R = and

= (FN=0)/

It was found that for starboard rudder depends on drift angle:

S = -0,34 + 0,41

For port side rudder variation of is symmetrical in relation to .

Leading edge of rudder is located close to propeller disk and it would be justified to apply the simple assumption of wR=wP. However, the preliminary simulations revealed that in present model separate approximation has to be applied for wake fraction at location of rudder. The normal force measured during oblique towing and the wake fraction in propeller disk calculated according to approximation shown in Fig.5 were used in determination of wake fraction wR. Developed approximation of wR is presented in Figures 6 and 7.

0,3 0,25 0,2 0,15 wR 0,1 0,05 0 -60 -40 -20 -0,05 -0,1 [deg] 0 20 40 60
= 16 deg

8 deg 0 deg -8 deg -16 deg

Figure 6. Approximation of wake fraction in location of starboard rudder

= -16 deg 0,3

0,25 0,2 0,15 w 0,1 0,05 -60 -40 0 -20 -0,05 0 -0,1 [deg] 20 40 60
wR wP

= 0 deg 0,3
0,25 0,2 0,15 w 0,1 0,05 -60 -40 0 -20 -0,05 0 -0,1 [deg] 20

wP

wR

40

60

= 16 deg 0,3
wP

0,25 0,2

wR

0,15 0,1 0,05

-60

-40

0 -20 -0,05 0 -0,1 [deg]

20

40

60

Figure 7. Comparison of propeller and rudder wake fraction

5. NUMERICAL SIMULATION OF EVASIVE ACTION MANOEUVRE The author considered the evasive action manoeuvres with rudder angles of 20 and 45deg and rudder executions started at given yaw rate (Administrative instruction No 1). In practice the results of that test are used to evaluate the controllability of inland waterway vessels. In present research the same procedure was used for assessment of mathematical model for numerical simulation of ship motion. Both model tests and simulations were carried out in shallow water of h/T=1.89, starting with approach speed of 0.773m/s that corresponds to 13km/h in full scale. End time of manoeuvre when yaw rate reaches zero after third execution of rudder, the magnitude that is a criterion in assessment of controllability, is compared in Table 3. Variation of yaw rate, rudder angle, thrust of starboard propeller and normal force on starboard rudder are illustrated in Figures 8 and 9, in comparison with values measured during free-sailing model test in towing tank.

Table 3. End time of evasive action manoeuvre

Rudder angle 20 deg 45 deg

Model test 8.4 s 10.7 s

Numerical simulation 7.4 s 10.4 s

25 20 15 10 [deg] 5 0 -5 0 -10 -15 -20 -25 t [s] r (test) 2 (sim.) (test) r (sim.) 4 6 8

2,5 2 1,5 1

4 3

T (sim.)

T (test) 2 T, FN [N]
r [deg/s]

0,5 0 10 -0,5 -1 -1,5 -2 -2,5

1 0 -1 -2 FN (sim.) -3 -4 t [s] 0 2 4 FN (test) 6 8 10

Figure 8. Evasive action manoeuvre with rudder angle 20deg, first rudder execution to starboard

50 40 30 20 (sim.) (test)

5 4 3 2
r [deg/s]

8 T (sim.) 6 4

[deg]

10 0 -10 0 -20 -30 -40 -50 5

1 0 10 r (sim.) r (test) 15 -1 -2 -3 -4 -5

T, FN [N]

2 FN (sim.) 0 -2 -4 -6 -8 t [s] 0 5 10 FN (test) 15 T (test)

t [s]

Figure 9. Evasive action manoeuvre with rudder angle 45deg, first rudder execution to starboard

ACKNOWLEDGEMENT The research reported in this paper was financially supported by the Minister of Science and Higher Education under grant No. N509 02932/2113.

REFERENCES Administrative taking evasive of instruction action 19 No 1: turning 2008 the Requirements capacity, relating In: to the capacity for

and

COMMISSION Directive down

DIRECTIVE of the

2008/126/EC European

December and of

amending laying

2006/87/EC technical

Parliament

Council

requirements

for inland waterway vessels, Official Journal of the European Union, L32/1 Bulletin of the SNAJ, No.668, February 1985, Prediction of manoeuvrability of a ship Kijima K., Katsuno K., Nakiri Y., Furukawa, Y., (1990), On the manoeuvring performance of a ship with the parameter of loading condition, Journal of the Society of Naval Architects of Japan, Vol.168

CONTACT DETAILS Tomasz Tabaczek, Wrocaw University of Technology, Wrocaw, Poland tomasz.tabaczek@pwr.wroc.pl

Manoeuvring Abilities of Podded Ships with Different Stern Shapes

Maciej Reichel Ship Design and Research Centre - CTO S.A., Gdask, Poland

For correct prediction of ship manoeuvring abilities, it is necessary to understand the influence of different parts of vessel on manoeuvring characteristics. In case of standard single propeller, single rudder ship, forces and interactions between hull, propeller and rudder can be quantitatively very well predicted. However, in the case of pod-driven ships, hull shape differs significantly from a propeller-driven ship and the interactions between hull and pod drive are not fully recognized. In this paper the manoeuvring model tests of pod-driven ship with different stern shapes are presented. During the tests a round bilge and a hard chine stern part, both with two different skeg lengths have been investigated. Standard manoeuvring tests have been performed and on this basis an attempt to describe the influence of the hull shape on manoeuvring characteristics has been made.

Keywords: manoeuvring abilities, pod-driven ships, stern shapes, model tests.

1. INTRODUCTION At the design stage it is very important to know or to predict the performance characteristic of a new ship. In almost all cases the most important issue is of course resistance and propulsive performance. But in some cases the most important are seakeeping abilities due to expected bad weather conditions or manoeuvring abilities due to long work time in port or in other restricted waters. In case of standard single propeller and single rudder ship, designers experience is sufficient for achievement good manoeuvring characteristics. However, in case of pod-driven ships, which of performance is still not fully recognised the designer experience may not be sufficient to guarantee a satisfactory manoeuvring characteristics. Therefore, it is necessary to assess the manoeuvring abilities of this type of ships with more reliable methods. Several methods can be chosen for the analysis. One of them is the application of Computer Fluid Dynamics CFD (Simonsen, Stern 2005). The computations quite often allow to calculate simultaneously hydrodynamic forces and to solve the equations of motions. The preparation of computational meshes and the time needed to reach convergence are long and therefore expensive. Another method for manoeuvring predictions is the simulation. In this method, the equations of motion are equipped with hydrodynamic derivatives and solved numerically (Kijima, Nakiri 2003). The derivatives can be obtained either from captive model tests or from empirical formulas. In case of performing the captive model tests it is necessary to build a physical model, what is time and money consuming. In case of empirical formulas parameters of the investigated ship should be in the range of the method, what is not always possible. The most reliable method for a certain ship is carrying out the model tests. This method is on one hand time and cost consuming, but on the other hand the detailed geometry of the ship, propeller and steering devices can be fully modelled and therefore results are more suitable for the specific ship. Results of the tests can be directly compared to requirements and modifications of rudder or propeller quickly introduced (Grski, Reichel 2007). In this paper the manoeuvring abilities of an LNG carrier with different stern shapes and with different skeg lengths are presented. For manoeuvrability assessment the free sailing model tests have been carried out. Results obtained from the tests have been compared between each other and against the IMO requirements. On the basis of this results the influence of different stern shapes and different skeg lengths has been determined.

2. SHIP AND MODEL DATA A 117.10-meter long LNG carrier has been chosen for this study. A wooden model was manufactured to a scale of 23.4. Two stern part shapes have been used for the analysis round-bilge (U) and hard-chine (L) type. The particulars of the models with different skeg lengths are shown in Table 1. Lateral area of skegs is shown in Table 2. Hull sections of both versions are shown in Figure 1. Hull shape from bow to midship was identical for both versions. The model was propelled by two azimuthing Z-drives, which were equipped with nozzles. Figure 2 presents a sketch of used propulsors. Details of propeller and azimuthing drives are presented in Table 3.

Table 1. Principal particulars of the models

Dimension Length Breadth Draught

CB

U-type

Small skeg

C Pa CWa

Large skeg

CB C Pa CWa

L-type 5.004 m 1.017 m 0.261 m 0.781 0.773 0.797 0.780 0.966 0.989 0.1675 0.0500 0.782 0.774 0.799 0.782 0.966 0.989 0.1692 0.0505

Table 2. Lateral area of skegs

Dimension AS AS / Ld

Small skeg 0.0353 m2 2.71 %

Large skeg 0.0589 m2 4.51 %

Table 3. Details of propellers and azimuthing drives

Dimension Number of blades Propeller diameter Pitch ratio at r/R=0.7 Expended blade area ratio Gondola length Gondola diameter Strut height (from propeller shaft)

4 148.31 mm 1.1102 0.791 142.20 mm 80.21 mm 112.00 mm

Figure 1. Hull sections of round-bilge stern with small skeg (left) and hard-chine stern with large skeg (middle)

Figure 2. Sketch of azimuthing drive used during the project

3. MANOEUVRING TESTS Standard IMO manoeuvring tests have been performed to determine the steering abilities and directional stability of the model with different stern shapes. Spiral tests, zig-zag tests and turning tests have been performed. All the model tests have been carried out in the test station Joniny at the lake Wdzydze. Approach speed for all tests was the same and equal to 0.85 m/s, what corresponds to 8 knots in full scale. Table 4 presents the detailed test programme, which have been performed for all versions.

Table 4. Model test parameters

Type of test zig-zag zig-zag turning spiral

Approach speed full scale model scale

Rudder angle

Heading deviation angle

U FS
8.0 kn

UM
0.85 m/s

20 10 35 -35 35

20 10 -

3.2 Model tests results On the basis of the model tests, the standard manoeuvring characteristics have been determined. Results of spiral tests are shown in Figures 4-7, from zig-zag tests in Figures 8-11 and from turning test in Figures 12, 13. Manoeuvring characteristics, such as advance, transfer, tactical diameter and turning radius, first and second overshoot angle and initial turning ability have been specified. Results have been checked with the IMO requirements and are shown in Tables 5-7.

Figure 4. Spiral test results for U-type with small skeg

Figure 5. Spiral test results for L-type with small skeg

Figure 6. Spiral test results for U-type with large skeg

Figure 7. Spiral test results for L-type with large skeg

Table 5. Instability loop parameters

Hull version U-type small skeg U-type large skeg L-type small skeg L-type large skeg

Loop height ~ 3.7 /s ~ 2.0 /s -

Loop width ~ 2.5 ~ 0.5 -

From the results of spiral tests it is clear that the hull with hard-chine stern part is directly stable, while the hull with round-bilge stern part is directly unstable. It could be also seen that increase on skeg area of 1.8 % Ld slightly increases the directional stability. The rates of turn are linear in the range of rudder deflection angle higher than 5. For rudder angle equal to 35 rate of turn for larger skeg is higher of about 20 % for round-bilge hull and of about 4 % for hard-chine hull. The directional stability was also proven by standard 10/10 and 20/20 zig-zag tests. Results of these tests are shown in figures below.

Figure 8. 10/10 zig-zag test results for both versions with small skeg

Figure 9. 20/20 zig-zag test results for both versions with small skeg

Figure 10. 10/10 zig-zag test results for both versions with large skeg

Figure 11. 20/20 zig-zag test results for both versions with large skeg

Table 6. Zig-zag manoeuvre characteristics

Hull version IMO criterion U-type small skeg U-type large skeg L-type small skeg L-type large skeg

First rudder deflection side PS SB PS SB PS SB PS SB

10/10 test First Second overshoot overshoot angle angle 10 25 5.6 16.1 13.1 10.3 4.6 10.4 6.7 7.5 4.4 9.5 5.5 8.7 3.6 10.1 4.4 7.8

20/20 First overshoot angle 25 12.0 13.9 9.4 10.8 10.8 10.1 8.5 9.7

Initial turning ability 2.5

1.5
1.0 1.3 1.1 1.4 1.1 1.4 1.3

After zig-zag tests it could be seen that the results are not in full compliance with spiral test, which shows directional instability of hull with round-bilge stern part. Both hull versions with both skeg lengths show good or very good directional stability during zig-zag tests. Only in one case, for round-bilge hull with small skeg the IMO criterion is not fulfilled. It could be seen that for hard-chine version the response on rudder deflection is quicker than for round-bilge version.

Figure 12. Turning test results for both versions with small skeg

Figure 13. Turning test results for both versions with large skeg

Table 7. Turning circle characteristics

Hull version IMO criterion U-type small skeg U-type large skeg L-type small skeg L-type large skeg

First rudder deflection side PS SB PS SB PS SB PS SB

Advance 4.50 1.86 1.75 2.02 1.81 1.83 1.76 1.98 1.78

Transfer

Tactical diameter 5.00 1.26 1.20 1.44 1.27 1.37 1.05 1.44 1.24

Steady turning diameter 0.82 0.71 0.77 0.66 0.78 0.62 0.74 0.61

0.44 0.47 0.58 0.50 0.56 0.37 0.67 0.52

Turning characteristics are in general similar to other pod-driven ships. Advance and tactical diameter fulfil IMO criterion. Steady turning diameter is for all cases smaller than the ship length. Minor differences appear in turning characteristics between hull with round-bilge and hard-chine stern shape.

4. CONCLUSIONS In this paper results of manoeuvring model tests of an LNG carrier are described. During the research two versions of stern part with two different skeg lengths have been tested. Standard manoeuvring model tests have been carried out to determine the dependence of manoeuvring abilities on the shape of stern part sections. Results of model tests show that no significant difference between two tested hull shapes can be observed. Both versions response on rudder deflection with similar force, nevertheless ship with hard-chine stern part responses faster than the one with round-bilge shape. This dependence is visible both for small skeg and for large skeg. Turning abilities of both hull versions are similar, but slightly better manoeuvring to starboard direction can be seen. Differences in advance between tested hulls with small skeg are at the level: in advance of 12 %, in transfer of about 2025 %, in tactical diameter of about 10% and with large skeg respectively 2 %, 10 % and 1 %. Steady turning diameter is smaller 510 % for hull with large skeg than with small skeg for both versions.

ACKNOWLEDGEMENTS Experiments described in this paper were a part of the research project N 509 040 32 / 2918 entitled Determination of hydrodynamic interactions between Z-drives and hull in manoeuvring conditions funded by Polish Ministry of Science and Higher Education.

REFERENCES Grski W., Reichel M., (2007), Free sailing model tests of evasive action manoeuvre of a river cargo motor barge on shallow water, Polish Maritime Research, No 2007/S2, pp. 95-97 Kijima K., Nakiri Y., (2003), On the Practical Prediction Method for Ship Manoeuvring Characteristics, MARSIM 2003, Terscheling, The Netherlands Simonsen C.D., Stern F., (2005), RANS Maneuvering Simulation of Esso Osaka with Rudder and a Body-Force Propeller, Journal of Ship Research, Vol.49, No.2, pp. 98-120

CONTACT DETAILS Maciej Reichel, Ship Design and Research Centre - CTO S.A., ul. Szczeciska 65, Gdask, Poland maciej.reichel@cto.gda.pl

Verication of various methods for calculation of diracted wave eld around the ship
Pawe Wroniszewski1 , Jan Jankowski1 , Stefan Grochowalski1
1

Polish Register of Shipping, Gdask, Poland

Diracted waves around a simple, prolongated ship hull were calculated using source methods and boundary element method, for two and three dimensional models. The comparison of numerical results and experimental data is presented, together with the discussion of numerical and experimental methods. Keywords: wave diraction, Boundary Element Method

INTRODUCTION

Water trapped on deck of small vessel can cause capsizing of the vessel, therefore, proper evaluation of amount of water on deck is substantial. It depends on the actual position of the upper edge of bulwark and on the position of diracted wave surface during the vessel motion in waves. One of the essential elements of the water on deck problem is determination of actual deformed wave surface which is the result of superposition of velocity potential of incident wave and velocity potential of diracted eld. As the velocity potential of undisturbed wave is given in a form of a function, it is enough to solve the diraction problem in order to determine the deformed wave elevation. The paper presents four methods of solving the diraction problem and validates them against a simplied shiplike body, for which appropriate experiments were carried out:

 Two and three dimensional wave source methods (2D WSM and 3D WSM); Two and three dimensional Rankine source methods (2D RSM and 3D RSM).
The results of computations for the experimental conditions by the four above methods are compared with the results of the experiments and discussed in the paper.

THE HYDRODYNAMIC BOUNDARY  VALUE PROBLEM DETERMINING THE DIFFRACTION

Neglecting viscous eects and assuming that water is incompressible and its ow irrotational, the diraction problem can be formulated in terms of the potential ow theory. Assuming additionally, that incident waves of small amplitude are diracted on the restrained body, the problem can be linearized. The assumption of linearity enables to superimpose the incident and diracted wave elds to obtain the approximation of real waves around the restrained body. The incident wave potential is given by the following formula:

g x W (x, t) = iA ekx3 ikik+i(t+ ) = W (x)ei(t+ ) ,

x R 3

(1)

where R = {x : x3 < 0} is a lower half space, x = (x1 , x2 , x3 ), x = (x1 , x2 , 0), k = 3

(k cos , k sin , 0) = (k1 , k2 , 0) is a wave vector, k =
2 g

is the wave number, is the

incident wave frequency, is the angle between the wave vector and xed vector in the body (normally, in the body symmetry plane, directed to the bow), and is the horizontal translation of the wave, in the direction of the wave vector. The elevation of the water surface is determined by the formula:

w (x, t) =

1 (x, t) g t

+ c ,
x3 =0

(2)

where potential = W + D is the sum of the potentials of the incident and diracted wave, and c is the vertical translation of the wave. It is assumed, that the diracted wave potential has similar form as wave potential:

D (x, t) = A (x) ei(t+ ) ,

(3)

where the time-independent part of the diracted potential (x) is determined by the following boundary-value problem (Jankowski, 2007):

Laplace equation

(x) = 0

x R \V, 3

(4)

where V is a closed domain occupied by the body, and the following boundary conditions:

On the free surface:

(x) k(x) = 0, x3 x SF = {x : x3 = 0} \V

(5)

On the wetted body:

D (x) W (x) = , n n x S0

(6)

where W is given by (1)

At innity:

 radiation condition:

lim

+ ik

= 0,

(7)

where =

(x1 y1 )2 + (x2 y2 )2 ;

 condition at innity on the free surface:

lim

|| c,

SF = {x : x3 = 0}

(8)

where c is constant;

 at the bottom:
x3

lim

= 0. x3

(9)

The solution to the problem is not simple and requires application of an approximate method to determine the diracted waves. Usually, the integral identities are used to solve the problem numerically, however, the fundamental solutions (or Green functions) must be given in explicit form.

FUNDAMENTAL SOLUTIONS OF THE BOUNDARY  VALUE PROBLEM

Depending on the method used to solve the problem (4) to (9) various fundamental solutions are used: 1. Fundamental solution in three dimensions (3D) satisfying Laplace equation, condition on the free surface and radiation condition and condition at innity on the free surface (Jankowski, 2007):
E(x, y) = 1 1 1 + G(x, z) , 4 |x y| |x z| x, y R , 3 x=y

(10)

where x = (x1 , x2 , x3 ), y = (y1 , y2 , y3 ) and z = (y1 , y2 , y3 )

|x y| = |x z| = (x1 y1 )2 + (x2 y2 )2 + (x3 y3 )2 (x1 y1 )2 + (x2 y2 )2 + (x3 + y3 )2

(11) (12)

and G is harmonic function in the integral form (Jankowski, 2007):

G(x, y) = 2 k |x z|

P.V. (J1 + iJ2 ) d ikek(x3 z3 )

eik d

(13)

where
= (x1 y1 ) cos + (x2 y2 ) sin P.V.J1 = eky [f cos kx + g sin kx] , P.V.J2 = eky [f sin kx g cos kx] ,

f (r, ) = + ln(kr) +
n=1 n

(kr)n cos n , n!n

r = 0,

g(r, ) = +

(kr) sin n , n!n n=1 = arctan (x3 z3 )

The rst term of function (10) represents the Rankin source, the second  its symmetrical mirror reection and the third  the pulsating source under free surface. 2. Fundamental solution in two dimensions (2D) satisfying Laplace equation, condition on the free surface and radiation condition and condition at innity on the free surface (Jankowski &Wyrzykowski, 1992):
E(x, y) = 1 ln r + ln r F (x, z) , 2

(14)

where r = |x y|, r = |x z|, x = (x2 , x3 ), y = (y2 , y3 ), z = (y2 , y3 ), and

F (x, z) = F R (x, z) + iF I (x, z), F (x, z) = 2ke
R k

(15)

[cos (k) (0.577215664 + ln (krp ) + C) + sin (k) ( + S)] ,

F I (x, z) = 2kek cos (k) ,

where
= x 2 y2 , = x 3 + y3 , rp = 2 + 2, ,

= arctan

C=
n=1

fn cos (n) , n fn sin (n) , n

S=
n=1

fn =

fn1 rp , n

f1 = 1.

3. Fundamental solution in 3D satisfying only Laplace equation (Marcinkowska, 1986):

E(x, y) = 1 1 , 4 |x y| x, y R , 3 x=y

(16)

where x = (x1 , x2 , x3 ), y = (y1 , y2 , y3 ). This is the rst term of function (10)  the so called Rankine source. 4. Fundamental solution in 2D satisfying only Laplace equation (Marcinkowska, 1986)
E(x, y) = 1 ln |x y| , 2 x, y R , 2 x=y

(17)

where x = (x2 , x3 ), y = (y2 , y3 ). It is the Rankine source in 2D domain. Thus the complexity of the fundamental solution depends on the number of boundary conditions the solution satises.

METHODS OF SOLVING THE HYDRODYNAMIC BOUNDARY  VALUE PROBLEM

In both methods presented below, the values of the potential in the computation domain are obtained by calculating the integrals over the boundary of the domain. The boundary is discretised, therefore the set of algebraic equations have to be solved. Such approach is in general called the Boundary Element Method.

Wave source method

In the case of using the fundamental solution (10) it is assumed that the solution to the boundary-value problem (4) to (9) takes the form of the following single layer potential:
(x) =
S0

(y)E(x, y)dsy

(18)

where is the source density function (complex function of real variables), and S0 is the wetted surface of the body (Fig. 1).

Figure 1: Wetted surface of the body

By substituting the singdile layer potential (18) to problem (4) to (9) the following Fredholm integral equation of second kind is obtained:
(x) + 2 (y)
S0

E (x, y) dsy = W (x) x n x n

(19)

where n is the normal vector to S0 , directed to the water. The discretisation and solving the equation (19) is done according to (Jankowski, 2007). The same approach is used to solve the problem using function (14), however, in this case the integral is over the cross section line of the cylindrical body (Fig. 2).

Rankine source method

In the case of using the fundamental solution in the form of (16) or (17) the following formula for harmonic functions, which represents the values in bounded domain by its

Figure 2: Cross section line

values and values of normal derivatives on the boundary  Fig. 3, (Marcinkowska, 1986), is used to solve the problem:
(x) =

(y)

E(x, y) dy y n

E(x, y)

(y) dy ; y n

(20)

normal vector is directed outside the domain. For 2D case the cross section of is used to determine the diraction potential.

Figure 3: Bounded domain in which diraction potential is determined, using fundamental solution (16, 17)

Equation (20) determines the potential inside the domain , unless the value of the potential and its derivatives on the boundary are known. Moving the point x into the boundary, the 'boundary-only' equation, that can be solve numerically, is obtained:
c(x)(x) =

(y)

E(x, y) dy y n

E(x, y)

(y) dy . y n

(21)

The function c(x) is caused by the singularity under the integrals (20), which appears when the point x is moved to the boundary. The singularity can be estimated by surrounding the boundary point x by a small sphere of radius , and calculating the integrals in equation (20) in the limit as
0. It can be shown (Marcinkowska (1986)), that for a smooth

boundary, for both 2D and 3D problems, the function c(x) is constant and equal to 0.5. To solve the equation (21) numerically, the boundary (surface) of the domain has to be discretised into N elements (panels) i , where
N i

i = (Fig. 3) and the equation

can be written as a sum of integrals. Each element has a node at point xi , located at the center of the element. Assuming that the value of the potential and its normal derivative is constant on each element and equal to the value in the node, they can be taken outside of the integrals. The set of equations for each element will consist of N equations of the form:
N

0.5i
j=1

j
j

E (xi , y) dy = y n

N j=1

j n

E (xi , y) dy .
j

(22)

where the lower indices i, j indicate, that the value was taken in nodes xi , xj respectively. The following notation is used:
aij =
j

E (xi , y) dy , y n E (xi , y) dy .

(23) (24)

bij =
j

If at elements are used for the discretisation, all elements aii are equal to zero. Then the set of equations can be written in a matrix form:
A = B , n 1 b11 b12 2 b21 b22 . = . . .. . . . . . . . N bN 1 bN 2

(25)
b1N b2N

0.5 a12 a21 0.5 . . .. . . . . . aN 1 aN 2

a1N a2N

. . .

. . .

1 n 2 n

. . .

(26)

0.5

bN N

N n

In fact, as the diraction potential is a complex value, there are two sets of equations. To solve them, boundary conditions (5) to (9) are introduced, in order to specify either
i or
i n .

The information provided by the boundary conditions can be in a form of a

proper value given to the unknown (6), or as the relation between the potential and its derivative. The condition on the free surface (5) brings no diculties, as it is identical with the normal derivation. As for the conditions at the innity (7) to (9), it has been assumed that the borders of the computational domain are far enough from the ship hull to directly specify the conditions, saying that the direction is identical with normal direction. It is assumed that the radiation condition (8) is automatically satised, what have been veried in the numerical results. The boundary conditions at the innity (7) bring the relation between the real and imaginary parts, therefore actually one has to deal with an extended set of 2N equations, which can be solved using algebraic methods.

DESCRIPTION OF THE EXPERIMENT

The results of captive model tests dedicated to measurement of volume of water shipping on deck, carried out at the Institute for Marine Dynamics, Canada, are used for validation purpose in this work. The tests in question were done in addition to experimental study of hydrodynamic forces generated on submerged part of deck in waves (Grochowalski, 1997). A cylindrical model with a constant cross-section in the form of semi-circle was used in the tests. The model was mounted rigidly to the tank carriage at various heading angles to oncoming waves and the tests were controlled and recorded on the carriage. Oscillations of water level at the model sides were measured by one wave probe at each model side (Port and Starboard). In addition, the undisturbed waves were measured far from the model place (Xcal) and at the side of the model arrangement (Starboard Far). The experimental setup is shown in Fig. 4 and 5.

Figure 4: Experimental model

The test parameters were: model draft, wave height, wave length and prole, heading angle and model drift velocity. The undisturbed wave prole and the water oscillations at both model sides were recorded continuously at various combination of experiment parameters.

Figure 5: Experiment setup for model tests of water shipping on deck

An example of recorded data is presented in Fig. 6. These results together with the simple form of the model constitute an excellent basis for a validation of theoretical models representing wave diraction and the numerical schemes for its computation.

Figure 6: Example run from the experiment

COMPARISON OF THE COMPUTATION RESULTS WITH THE EXPERIMENTAL DATA

The numerical calculations were performed for the case of a beam wave, approaching the ship hull under the angle of 270o (dened as in Fig. 5) and zero model drift velocity. Such case is easy to simplify from 3D to 2D, therefore the boundary element method and single layer potential method for both cases were used. Comparison of experimental data with computation results requires the proper incident wave potential, which has been obtained from the records of undisturbed wave (Xcal). The wave was separated into harmonics using discrete Fourier transform, and translated by the known distance (compare function (1)). The comparison of the undisturbed wave recorded far before the model (Xcal) and the wave in the area close to the model (SF in Fig. 5) is presented in Fig. 7. It should be taken into account that the undesirable inuence of the walls of the towing tank could have taken place, as the water level measured by the probe on the side of the

Figure 7: Example correlation between the translated Xcal wave used for calculations and the Starbord Far (SF) wave

model (SF probe) seems to be banked up. This eect seems to be not only due to the diracted wave originating in the model. The following graphs (Figures 8 to 11) present examples of comparison between numerical results and the experimental data.

Figure 8: Experiment No. 38 - nominal wave period T = 1.8s, nominal amplitude A = 0.19m, model draft: d = 0.1m, Port (weather side)

Figure 9: Experiment No. 38 - nominal wave period T = 1.8s, nominal amplitude A = 0.19m, model draft: d = 0.1m, Starboard (lee side)

Figure 10: Experiment No. 41 - nominal wave period T = 1.8s, nominal amplitude A = 0.19m, model draft: d = 0.04m, Port (weather side)

Figure 11: Experiment No. 41 - nominal wave period T = 1.8s, nominal amplitude A = 0.19m, model draft: d = 0.04m, Starboard (lee side)

There is essential eierence in the wave surface between the weather and the lee side. Large portion of the wave prole is damped by the presence of the body and the wave height at the lee side is signicantly reduced. The theoretical methods reect this dierence to certain degree. The theoretical methods presented here represent very well the deformation of waves at weather side, and all the methods give the results very close to each other. In case of the lee side, the theoretical methods dier between themselves and with the experimental results. The computation results which are the closest to the experiments are provided by the 3D Rankine source method (RSM 3D). Although the correlation between the numerical calculation and the experiment data is not perfect, the better correlation is hardly to be expected by the linear, potential methods. The order of magnitude of the numerical results ts well into the experimental data for the weather side, while the nonlinearities could not be reected well from the obvious reasons, in particular for the lee side.

CONCLUSIONS

The denition of the radiation conditions in the wave diraction problem for use by the Rankine source method (RSM) is nontrivial, therefore one of the aims of this work was to verify if the approach provides relevant results. The results of the Rankine source method correlate well with the results of the wave source method (WSM) for the weather side. It means that the diracted wave could freely leave the domain, without any reection on the outer boundaries. The correlation on the lee side is not sucient, and it requires further study. Both methods have been proved usable. In the Rankine source method the system of algebraic equations that has to be solved is much bigger, as the equations are dened not only for the panels at the ship hull, but also for the mesh generated for the water surface, side-borders and lower-border of the computational domain. After calculations, the potential for each boundary element is given. In case of the wave source method, the system of equations is smaller, but computation of the potential in each point requires more time, as complicated series have to be calculated. These features make the Rankine source method ecient for determination of whole diracted wave eld around the ship, while the wave source method seems to be better for computing the value of the potential in the required points in space. The inuence of the density of the computational mesh on the results should be studied, especially of the density of the free surface mesh in the Rankine source method. The nonlinear eects seem to be signicant in determining the free surface around the ship. Further studies of this topic should be continued.

REFERENCES
Grochowalski S., (1997), 'Experimental Investigation of the Hydrodynamic Forces and Pressure Distribution on the Submerged Part of Deck in Waves', Vol.1 Technical Report, National Research Council Canada, Institute for Marine Dynamics, TR-1997-09. Jankowski J., (2007), 'Ship facing waves', Technical Report No. 52, Polish Register of Shipping, Gdask, Poland (in Polish). Jankowski J., Wyrzykowski J., (1992), 'Continuation of Haskind's works', Transactions of the First Int. Conference in Commemoration of the Anniversary of Creating Russian Fleet by Peter the Great, St. Petersburg. Marcinkowska H., (1986), Introduction to the partial dierential equations, Wydawnictwo Naukowe PWN, Warsaw.

CONTACT DETAILS
Pawe Wroniszewski, Polish Register of Shipping, Hallera 126, 80-416 Gdask, Poland pwroniszewski@prs.pl

Ducted and Open Propeller Subjected to Intermittent Ventilation

Anna M. Kozlowska1,2, Sverre Steen1,2
1

Department of Marine Technology, Norwegian University of Science and Technology, Trondheim, Norway
2

Rolls Royce University Technology Centre Performance on a Seaway, Trondheim, Norway

The analyses in this paper are based on an experimental study performed by Koushan (2006 I, II, III) and also based on a new set of experiments. The main motivation for the new experiments was to monitor the time for ventilation to be established. The use of a 6 component force dynamometer on one of the four propeller blades gives detailed insight into the forces on propeller, while the use of high speed of underwater video gives a visual understanding of the ventilation phenomena. In this large experimental campaign an azimuth thruster with both open pulling and ducted pushing propeller were tested. Propeller immersion ratios, carriage speed, propeller rate of revolutions, azimuth angles, period of heave oscillations and oscillations amplitude were varied during the experiments. Analysis of the propeller ventilation can be divided in two parts: comparison between ventilation in static and dynamic conditions (heave motion) both for open and ducted propeller and the discussion how to estimate thrust loss. As a conclusion, a new formulation of the relations between ventilation and thrust losses was developed.

Keywords: Intermittent ventilation, azimuth thrusters, thrust loss, torque variations

SYMBOLS INDEX
A0
Anv Av cL cLC D, R

[m 2 ]
[m 2 ] [m 2 ]

propeller disc area non ventilated propeller disc area ventilated propeller disc area lift coefficient lift coefficient for partially ventilated propeller propeller diameter, propeller radius blade area ratio, free surface propeller submergence advance ratio nominal torque coefficient total torque coefficient nominal thrust coefficient total thrust coefficient propeller revolutions

pat
pcav Qn Qt Tn Tt V V

[ Pa ] [ Pa ] [ Nm] [ Nm] [N ] [N ] [m / s] [m / s]

[-] [-]
[ m]

EAR

FS
h J KQ n

[-] [-]
[ m]

[-] [-] [-] [-] [-]

[ Hz ]

z T ,
Q V VC
V

[-]
[deg]

[-] [-]

atmospheric pressure pressure in cavity nominal torque (no ventilation) total torque nominal thrust (no ventilation) total thrust advance speed velocity seen by propeller blade number of blades angle of attack total thrust/torque loss factors

K Qt KT n KT t
n

thrust losses for fully ventilated propeller thrust losses for ventilated [-] propeller [kg / m3 ] density of water [-] ventilation number

1. INTRODUCTION. Ventilation can occur both in low speed operations like Dynamic Positioning (DP), and in transit conditions. Ventilation typically occurs when the propeller loading is high and the propeller submergence is limited, and when the relative motions at the propeller are large due to heavy seas (Smogeli 2006). Ventilation leads to a sudden large loss of propeller thrust, which might lead to propeller racing and possibly damaging dynamic loads, as well as noise and vibration (Koushan 2006). It seems that more knowledge is needed to understand the typical magnitude of thrust loss and dynamic forces related to the different types of ventilation. The overall goal of this work is to get a firm understanding of the fluid load mechanisms resulting in large and rapid changes in the propeller loading and also showing how to avoid the more harmful types of propeller ventilation, either by design requirements or by operational guidance for ships when operating in off design conditions. Ventilation have been studied experimentally as well as theoretically by several researchers since 1930 see e.g. Kempf (1930), Shiba (1953), Fleisher (1973), Faltinsen (1981, 1983), Minsaas (1975, 1981, 1983), Olofsson (1996), Koushan (2006) and Kozlowska (2009).

Kempf (1930) was one of the first researchers who studied the ventilation effect on propellers. He studied the torque and thrust losses due to ventilation using different propellers (three and four bladed) as well as different immersion ratios and propeller rate of revolutions. The ventilation effect on thrust and torque was also study by Shiba (1953), who discussed the influence of different propeller design parameters e.g. expanded area ratio, contour of blade, radial variation of pitch, skewback, effect of rudder, turbulence of original flow as well as scale effects on ventilation. Ventilation effects with respect to vessel operation in addition to added resistance in waves and reduction of propulsive efficiency can be found in Faltinsen (1981, 1983) and Minsaas (1975, 1981, and 1983). Olofsson (1996) studied the force and flow characteristics of surface piercing propellers. Koushan (2006) performed extensive model tests on an azimuth thruster with 6 DoF measurements of forces on one of the four blades on an azimuthing thruster, as reported in three papers (Koushan 2006 I, II and III). Koushan (2006 I) described the dynamics of ventilated propeller blade axial force on pulling thruster at bollard condition running at several constant immersion ratios and constant propeller rate of revolution. Koushan (2006 II) presented the dynamics of ventilated propeller blade axial force on pulling thruster at bollard condition and constant propeller rate of revolution moving with force sinusoidal heave motion. Koushan (2006 III) presented the dynamics of ventilated propeller blade and duct loadings at bollard condition and constant propeller rate of revolution. The data reported by Koushan (2006, I, II, III) are further studied and reported in this paper. Kozlowska (2009) focused on ventilation inception mechanisms, classification of types of ventilation, thrust loss related to each type of ventilation, and a simple calculation method for predicting thrust loss.

2. TEST SET UP AND INSTRUMENTATION. The experiments by Koushan (Koushan 2006 I, II, III) were conducted in Marine Cybernetics Laboratories at the Norwegian University of Science and Technology. This tank is 40 m long, 6.45 m wide and 1.5 m deep. The same propeller model was used for open pulling and ducted pushing configuration. In order to capture the dynamics a high sampling frequency was used during experiments. The propeller has a diameter of 250 mm, design pitch ratio of 1.1 and expanded area ratio of 0.595. Blade axial, radial, tangential forces and moments about all three axes were measured during the experiments. A pulse meter provided propeller rate of

revolutions as well as indicating the angular position of the reference blade. For more detailed description see Koushan (2006 I, II, III) or Kozlowska (2009). A new set of experiments was performed by Kozlowska and Califano (April 2009). The main motivation for these experiments was to monitor the time for ventilation to be established. These experiments were also performed in Marine Cybernetics Laboratories and the same open pulling propeller configuration was used. A novel blade dynamometer capable to

measure 5 degrees of freedom was used during these experiments. Tested conditions presented in this paper are summarized in table below:
Table 1. Test conditions for open and ducted propeller, static conditions.

Open propeller Koushan (2006 I) -0.26 0 0.5 1.0 1.72 2.6 3.4 h/R [-] static April 2009 1.6 1.68 1.76 1.84 1.92 2

Ducted propeller Koushan (2006 III) 0 1 2 2.4

Figure 1. Submergence ratios for open propeller

Table 2. Test conditions for open and ducted propeller with dynamic-heave motions.

Open propeller Dynamic submergences heave motions h/R [-] Amplitude/R Period min max [-] [s] -0.25 2.15 2.15 2 1.1 3.25 2.15 2 2.2 3.3 1.1 2 -0.25 2.15 2.15 4 1.1 3.25 2.15 4 2.2 3.3 1.1 4

Ducted propeller h/R [-] min -0.92 0.08 0.93 -0.92 0.08 0.93 Amplitude/R [-] 1.8 1.8 1.1 1.8 1.8 1.1 Period [s] 2 2 2 4 4 4

max 0.88 1.88 2,03 0.88 1.88 2,03

The thrust loss factor is defined as follows:

T = Tt Tt = Tn KTn n 2 D 4

TT total thrust, Tn nominal thrust (no ventilation)

(1)

T = 1 means no thrust loss and T = 0 means total thrust loss due to ventilation
and out of water effects.

3. VENTILATION INCEPTION. In Figures 2 and 3 the ventilation inception time is plotted as a function of propeller revolutions, submergences and forward speed. We tested three propeller revolution n=12Hz, 14Hz, 16Hz at three different carriage speeds v=0.3m/s, 0.35m/s, 0.45m/s which means that we have the same loading parameter (advance speed) for different revolutions and carriage speeds The test conditions are given in Table 3 below.
Table 3. Ventilation inception, test conditions.

n=12 Hz 0.3 0.35 0.4

n=14 Hz n=16 Hz Carriage speed [m/s] 0.3 0.3 0.35 0.35 0.4 0.4

n=12 Hz

n=14 Hz n=16 Hz advance speed J[-] 0.086 0.100 0.114 0.075 0.088 0.100

0.100 0.117 0.133

Figure 2. Ventilation inception time as a function of propeller submergence and propeller speed (J=0.1 for n=12,14,16 Hz)

Figure 3. Ventilation inception time as a function of propeller submergence and forward speed (J=0.1 for n=12,14,16 Hz)

From the result it is found that the ventilation inception time increases by decreasing the propeller speed or by increasing the carriage speed.

4. THRUST HYSTERESIS EFFECT. The hysteresis effect on thrust for an open propeller is presented in figure 4. The figure shows the time series of thrust as a function of submergence ratio h/R during two different ventilation events. Ventilation is generated by sinusoidal vertical motion of the propeller with different amplitudes. For h/R>3 the propeller is deeply submerged and we observe no thrust loss. When the submergence is reduced full thrust is maintained until h/R<1.5 where ventilation starts and then the thrust drops to 23% of nominal thrust at h/R=1.

For h/R<1 thrust loss is mainly due to loss of propeller disc area and we observe no hysteresis effect. When the submergence is increased the thrust is built up until we achieve 100% of nominal thrust for h/R=3. A similar phenomenon is also observed for the ducted propeller, see figure 5. So it seems that the thrust loss depends on the direction of motion when the propeller is going down the thrust loss is bigger compared to upward motions, which supports the conclusions that this is the time-dependence of the ventilation phenomenon that causes most of the hysteresis effect.

Figure 4. Hysteresis effect, open propeller Dynamic heave motion (T=4s)

Figure 5. Hysteresis effect, ducted propeller Dynamic heave motion (T=4s)

Figure 6 and figure 7 show that the thrust loss coefficients for open and ducted propellers in static condition (no heave motion). It is observed from the figures that also in static condition the thrust loss shows something similar to a hysteresis effect, since for the same value of submergence and propeller loading the propeller can both be ventilated and non-ventilated, resulting in significantly different thrust loss values. This is related to the fact than it takes time for ventilation to be established in well submerged conditions, just as for the dynamic heave motion conditions.

Figure 6. Thrust loss coefficient for open propeller in static condition (no heave motion), bollard condition.

Figure 7.Thrust loss coefficients for open propeller in static condition (no heave motion).

5. COMPARISON BETWEEN STATIC AND DYNAMIC THRUST LOSS (HEAVE MOTION). In order to check the importance of dynamic heave motion for thrust loss due to ventilation, the thrust loss as a function of submergence ratio is plotted in dynamic and static conditions for both open and ducted propellers. It is seen from Figure 8 than for open propeller, the thrust is always larger in condition with dynamic heave motion than in static condition. From Figure 9 it is seen that for a ducted propeller there is less difference between thrust in static and dynamic conditions, although thrust in dynamic heave is still slightly larger than in static condition. The difference between dynamic and static thrust loss due to heave motion can be 20% for deeply submerged propellers in bollard conditions. The effect of dynamic heave motion becomes less important for higher advance ratios

Figure 8. Open propeller, bollard condition. Propeller speed n=14Hz, Period of heave motion: 4 s

Figure 9. Ducted propeller, bollard condition. Propeller speed n=14Hz, Period of heave motion: 4 s

6. IMPORTANCE OF THE HEAVE MOTION PERIOD FOR THRUST LOSS. In order to check the importance of the heave oscillation period two different periods: 2s, 4s were tested. Based on the measurement result we found that we get more ventilation for longer heave amplitude. This can be also explained as the effect of the ventilation time dependent phenomena. For large heave amplitude the time is large enough for ventilation to be established. We can also observe the tendency that the heave motion period has not a significant influence for propeller submergence h/R<1, it is much more important for deep submerged propellers but still the difference in thrust loss is not too large and probably can be neglected.

Comparison between different heave periods (T=4s and T=2s)

Figure 10. Open propeller, v=0 m/s, n=14Hz

Figure 11. Ducted propeller, v=0 m/s, n=14Hz

7. COMPARISON BETWEEN THRUST LOSS OF OPEN AND DUCTED PROPELLERS. In order to have the comparable cases we compare open and ducted propeller in bollard conditions as well as open pulling propeller in pushing conditions (negative advance speed) with ducted pushing propeller.

Figure 12. Comparison between thrust loss for open and ducted propeller for bollard conditions, n=14 Hz, static conditions- no heave motion

Figure 13. Comparison between thrust loss for open pulling (v=-0.35 m/s) and ducted pushing propeller (v=0.35 m/s), static conditions no heave motion.

Figure 14. Comparison between open and ducted propeller for bollard conditions, n=14 Hz, dynamic conditions-heave motion (T=4s)

Figure 15. Comparison between open pulling: and ducted pushing propeller, dynamic conditions heave motion (T=4s).

As a conclusion we can say that ducted propeller ventilates later than the open propeller but if it does ventilate the thrust loss is larger for the ducted propeller. Less difference between open and ducted propeller is observed for dynamic heave motion than for static condition.

8. IMPROVED FORMULA FOR CALUCLATION OF THRUST LOSS. In Kozlowska (2009) a simple formula based on the fully ventilated propeller assumption and out of water effect from (Minsaas 1983) for predicting thrust loss was presented. The conclusion was that the formula over-predicted the thrust loss for a deeply submerged propeller. The reason for this is not clear, but it seems likely that the thrust loss factor can be a function also of the propeller loading, advance velocity and numbers of revolutions not only the submergence ratio. Also, for large submergence the propeller was not fully ventilated.

Figure 16. Comparison between calculations and experiments for low advance speed.

Figure 17. Comparison between calculations and experiments for bollard conditions.

The simple formula to predict thrust loss is based on the assumption that thrust loss is mainly due to loss of propeller disc area, wave making by propeller (if it works near the free surface), Wagner effect and ventilation. The Wagner effect (Wagner, 1925) is the effect that at sudden

immersion of a foil or propeller blade, the lift is just half of the steady lift. The lift grows (rapidly) with time, and converges asymptotically towards the steady lift. The growth of the lift is expressed as a function of the number of chord lengths traveled after immersion. It was shown that we get fairly good agreement between calculations and experiments for h/R<1.6. For deep submergence the current formula overpredicts thrust loss compare with experiments. A possible reason for this could be that for the deeply submerge propeller is not fully ventilated as it was assumed but only partially ventilated. In this paper we propose corrections based on the assumption that thrust loss depends on how much blade area is covered by air. This was verified based on visual observations using high speed video during the experiments. Figure 18 present the photographs which help to estimates how much blade
area is covered by air due to ventilation with respect to time for bollard condition.

h/R=2.6, Time=X rev. ( Av / A0 ) =0

h/R=2.6, Time=X+23 rev ( Av / A0 ) =0.15

h/R=2.6, Time=X+26 rev ( Av / A0 ) =0.6

h/R=2.6, Time=X+28 rev ( Av / A0 ) =0.7

h/R=1.4, Time=X rev ( Av / A0 ) =1

h/R=1.4, Time=X+50 rev h/R=1, Time=X rev ( Av / A0 ) =1 ( Av / A0 ) =1 Figure 18. ventilation, n=14 Hz, bollard condition

h/R=1, Time=X +30 rev ( Av / A0 ) =1

8.1 Corrections for 1.6<h/R<deep submergence For fully ventilated propellers the lift can be expressed by using an analogy to the lift of a cavitating foil by assuming that suction side of the propeller blade section is covered by air and pressure on the pressure side of the propeller blade section is equal to the static pressure.
cLV = cL (V = 0) + V

(2)

Where cL (V = 0) is the pressure side contribution at the ventilation number equal to zero which corresponds to fully ventilated case. By assuming that pressure side of the profile has no camber the lift coefficient is expressed as follows
cL (V = 0) =
2

(3)

Due to pressure difference between the static pressure on the pressure side and the atmospheric pressure on the suction side we get the lift coefficient (excluding effect of camber and angle of incidence) as follows:
V =

2 gh V 2

(4)

Finally the lift coefficient for a fully ventilated propeller blade section can be estimated by:
cLV = cL (V = 0) +

2 gh 2 gh = + 2 2 V2 V

(5)

The thrust coefficient of a typical propeller can roughly be approximated by:

K T = 1.5 EAR cL (0.7)

(6)

By combination of equation (5) and (6), the thrust loss for a fully ventilated propeller may be approximated as follows:
V =

1.5 EAR 2 gh ( + ) 2 KTn V

(7)

The lift coefficient for non ventilated regime can be approximated by:
cL (0.7) = K Tn 1.5 EAR

(8)

The thrust loss for a partly ventilated propeller depends on how much area of propeller blade is cover by air. The lift coefficient for partilally ventilated propeller might be approximated

from the formulas for lift coefficient of a non- ventilated flat plate and a fully ventilated flat plate, weighted by the ratios of ventilated and non ventilated areas.
cLC (0.7) =

KTn 2 gh ( Anv / A0 ) + ( + 2 ) ( Av / A0 ) 1.5 EAR 2 V

(9)

Thrust loss coefficient for partially ventilated propeller can then be calculated as follows:
VC = K 1.5 EAR 2 gh ( + ) ( Av / A0 ) + ( Tn ( Anv / A0 ) = KTn V EAR 2

1.5 EAR 2 gh ) ( Av / A0 ) + ( Anv / A0 ) = ( + 2 K Tn V

(10)

VC =1 means no thrust loss and VC =0 means total thrust loss due to ventilation.

where : Anv - non ventilated propeller disc area, Av - ventilated propeller disc area
A0 - blade area if Anv / A0 =1 it means that propeller doe not ventilate, if

Av / A0 =1it means that propeller is fully ventilated.

In the table below we summarize thrust loss calculations based on the assumptions that thrust loss due to ventilation depends on how much of the blade area is covered by ventilation. Based on visual observations using high speed video during the experiments we estimated how much blade area (minimum and maximum value) is covered by air during a ventilation event. For example for h/R=2.6, n=14Hz and bollard condition we estimated that the fraction of blade area ratio covered by air oscillated from 0 to 80% of total blade area. It means that thrust loss due to ventilation can oscillate from no thrust loss to 35% loss of thrust. Figure 19 shows comparison between the new formula for thrust loss (equation10) and experiments for a wider range of submergence ratios. It is seen that agreement is very good.

Table 4.Thrust loss calculations based on new formula for bollard conditions, different submergence ratios.

h/R [-] 1 1.4 1.72 2.6 3.4

( Anv / A0 )
min - max 0 0 - 0.05 0 - 0.5 0.2 - 1 0.2 - 1

( Av / A0 )
min - max 1 1 - 0.95 1 - 0.5 0 - 0.8 0 - 0.8

VC
min - max 0.3 0.32 - 0.4 0.405 - 0.68 0.65 - 1 0.68 - 1

Figure19. Comparison between calculated and measured thrust loss for an open propeller in bollard condition. Propeller speed 14 Hz

9. CONCLUSIONS New analysis has been done based on a series of experiments performed by Koushan (2006 I II, III) and a new set of experiments carried out by the authors in April 2009. The influence of several factors on ventilation and thrust loss were investigated. Based on the new analysis the following conclusions have been drawn: 1. It was shown that several factors i.e. type of propeller (open and ducted), advance ratios, propeller revolutions, submergences and time have influence on ventilation inception and thrust drop due to ventilation. 2. Thrust loss in static conditions is larger than the thrust loss measured during dynamic heave motion, especially for bollard condition, and more for open than for ducted propeller. The dynamic effect is less important for higher advance ratios. 3. For static condition ducted propeller ventilates later than open propeller but the thrust loss is higher when ventilation occurs. 4. A new formula for calculation of thrust loss is given. The formula relies on knowledge of the area of the propeller blades cover by air. The agreement with experimental results is very good.

ACKNOWLEDGEMENTS This work has been carried out at the University Technology Centre of Rolls Royce at NTNU as a part of SeaPro project, sponsored by the Rolls Royce Marine and the Research Council of Norway. The authors gratefully acknowledge the Rolls Royce University Technology Centre in Trondheim for supporting the project and Dr. K. Koushan for providing experimental data.

REFERENCES Faltinsen, O., Minsaas, K., Liapias, N., Skjrdal, S.O. (1981). `Prediction of Resistance and Propulsion of a Ship in a Seaway`. Proceedings of 13th Symposium on Naval Hydrodynamics, Edited by T.Inui, The Shipbuilding Research Association in Japan. Fleischer, K.P. (1973). Untersuchungen ber das Zusammenwirken von Schiff und Propeller bei tilgetauchten Propellern, Publication 35/75 of Forschungszentrum des Deutchen Schifbaus, Hamburg, Germany. Kempf, G. (1933). Immersion of Propeller, T.NEC, Vol. 50. Koushan, K. (2006 I). Dynamics of Ventilated Propeller Blade Loading on Thrusters, World Maritime Technology Conference, London, U.K. Koushan, K. (2006 II). Dynamics of Ventilated Propeller Blade Loading on Thrusters Due to Forced Sinusoidal Heave Motion, 26th Symp. On Naval Hydrodynamics Rome, Italy. Koushan, K. (2006 III). Dynamics of Propeller Blade and Duct Loadings on Ventilated Ducted Thrusters Operating at Zero Speed, Proceedings of T-Pod conference 2006, Conference held at L'ABER WRAC'H. Kozlowska A. M. , Steen S., Koushan K. (2009) Classification of Different Type of Propeller Ventilation and Ventilation Inception Mechanisms SMP 09, Trondheim, Norway. Minsaas, K., Wermter, R., and Hansen, A.G. (1975). Scale Effects on Propulsion Factors, 14 th International Towing Tank Conferences, Proceedings Volume 3. Minsaas, K., Faltinsen O., Person, B. (1983). On the Importance of Added Resistance, Propeller Immersion and Propeller Ventilation for Large Ships in a Seaway Proc. of 2 nd Int. Symp. on Practical Design in Shipbuilding (PRADS) ,Tokyo& Seul. pp.149-159. Smogeli Q.N. (2006). Control of Marine Propellers From Normal to Extreme Conditions , Ph.D. thesis, Department of Marine Technology , Trondheim.

Olofsson N. (1996). Force and Flow Characteristic of a Partially Submerged Propeller, Ph.D. thesis, Chalmers University of Technology, Goteborg. Shiba, H. (1953). Air - Drawing of Marine Propellers Transportation Technical Research Institute, Report no. 9, Japan Wagner, H. (1925). ber die Entstehung des dynamischen Auftriebes vom Tragflgeln Z.f.a.MM, Vol.5.Heft 1. Feb. 1925, pp.17-35.

CONTACT DETAILS Anna Maria Kozlowska, Department of Marine Technology at Norwegian University of Science and Technology, Otto Nielsens vei 10, Trondheim, Norway. anna.kozlowska@ntnu.no

Using the vortex generators for improving the wake flow of large ships - preliminary results of CFD analyses
Pawe Dymarski, Marek Kraskowski Centrum Techniki Okrtowej S.A., Gdask, Poland

Current research work of CTO is focused on the application of vortex generators for increasing the ships propulsion efficiency, by considerable improving the propeller inflow at the cost of additional resistance of the vortex generators. The total resistance of the vessel with vortex generators is not necessarily increased, as presented in the paper. The idea of vortex generators is based on that from aerospace industry, where the vortex generators are used to prevent the flow separation on the suction side of the airplanes wing the vortices introduced to the boundary layer are attaching to the wing surface. It is expected that it is also possible to take advantage of this effect in ship design, by reducing the boundary layer thickness and reducing the areas of separated flows in the propeller disc region. The preliminary investigation of the effectiveness of the vortex generators was carried out for a car carrier and a large bulk carrier. .

Keywords: wake flow, RANSE, vortex generators

1. INTRODUCTION The paper presents the results of preliminary investigation of the effectiveness of using the vortex generators for improving the wake flow of large merchant ships. The idea of vortex generators is based on that from aerospace industry, where the vortex generators are used to prevent the flow separation on the suction side of the airplanes wing the vortices introduced to the boundary layer are attaching to the wing surface. It is expected that it is also possible to take advantage of this effect in ship design, by reducing the boundary layer thickness and reducing the areas of separated flows in the propeller disc region. Some promising results were already published (Schmde, 2008) The vortex generators are just small winglets, located (in case of a ship) in the aft part of the hull, e.g. on the skeg, as presented in Figure 1. It is important that these winglets do not exceed the basic line and maximum breadth of the hull, to minimize the damage risk.

Figure 1 Vortex generators fitted to the ship hull sample.

The purpose of vortex generators is to improve the propulsion efficiency of the hull, by increasing the average flow velocity in the propeller disc and improving the uniformity of the wake flow. The presence of vortex generators introduce some additional resistance, however, the total resistance of the hull with generators is not necessarily increased, because in some cases the generators improve the pressure distribution on the aft part (e.g. by reducing the flow separation area). The results of CFD analyses for two large merchant vessels with vortex generators in different configurations are presented below. The computations were carried out at model scale, to enable direct comparison of the CFD results with the experimental results.

2. DESCRIPTION OF THE COMPUTATIONAL CASES In this chapter, main data of the analysed vessels as well as the geometry and tested configurations of the vortex generators are presented. Selection of the vessels was based on the measured wake flows both vessels are characterized with complex wake flow, troublesome in respect of the propeller design (high non-uniformity, strong vortices and low average axial velocity). It was expected that there is a large area for improvement for both of them.

2.1 Main data of the vessels Main characteristics of the hull shapes, scale factors and visualizations of the hull shapes are presented below.
Table 1 Main data of the vessels

Vessel Length b.p. [m] Breadth [m] Draught [m] Block coefficient [ - ] Speed [kn]

Car carrier 188.00 32.23 9.00 0.582 20.0

Bulk carrier 250.80 44.40 13.00 0.836 15.0

As mentioned above, the computations were carried out at model scale. The scale factors are the same as for the models used for towing tank experiments, and are listed in Table 2. The scaled values off speed are also given
Table 2 Scale factors and speeds in model scale

Vessel Scale factor Model speed [m/s]

Car carrier 26.60 1.995

Bulk carrier 38.00 1.252

Figure 2 Geometry of the hulls bulk carrier (left) and car carrier (right)

2.2 Geometry and configuration of the vortex generators In the presented analysis, the vortex generators were used in the arrays of two or three. The size of the generators was different, but the shape and proportions were the same for each of them see Figure 3.
b

Figure 3 Geometry of the vortex generator outline and profile

The proportions of the outline dimensions are: b = 1 , a = 2 . The dimension a is further h h used as a characteristic dimension of the vortex generator. Configurations of the vortex generators for two analysed vessels is presented in figures 4 8. Each configuration is described by the following numbers: Characteristic dimension a in model scale and in full scale Angle of attack, relative to the streamlines direction; positive angle of attack means that the streamlines are deflected upwards. The descriptions are given in tables 3 7. The order of numbering of the vortex generators is from stern to bow.

Figure 4 Vortex generators for the car carrier configuration 1 Table 3 Description of configuration 1 of vortex generators for car carrier

Base size a

Base size a

- model scale [m]

Generator 1 Generator 2 Generator 3 0.040 0.040 0.040

- full scale [m]

1.064 1.064 1.064

Angle of attack [deg] 15 15 15

Figure 5 Vortex generators for the car carrier configuration 2 Table 4 Description of configuration 2 of vortex generators for car carrier

Base size a

Base size a

- model scale [m]

Generator 1 Generator 2 Generator 3 0.030 0.030 0.040

- full scale [m]

0.798 0.798 1.064

Angle of attack [deg] 15 15 15

Figure 6 Vortex generators for the car carrier configuration 3 Table 5 Description of configuration 3 of vortex generators for car carrier

Base size a

Base size a

- model scale [m]

Generator 1 Generator 2 0.030 0.040

- full scale [m]

0.798 1.064

Angle of attack [deg] -15 15

Figure 7 Vortex generators for the car carrier configuration 1 Table 6 Description of configuration 1 of vortex generators for bulk carrier

Base size a

Base size a

- model scale [m]

Generator 1 Generator 2 Generator 3 0.040 0.040 0.040

- full scale [m]

1.520 1.520 1.520

Angle of attack [deg] 15 15 15

Figure 8 Vortex generators for the car carrier configuration 2 Table 7 Description of configuration 2 of vortex generators for bulk carrier

Base size a

Base size a

- model scale [m]

Generator 1 Generator 2 Generator 3 0.020 0.030 0.040

- full scale [m]

0.760 1.140 1.520

Angle of attack [deg] -15 -15 15

3. METHODOLOGY OF THE COMPUTATIONS The CFD computations of the flow around the ship hulls with vortex generators were carried out with the use of RANSE flow model, implemented in STAR CCM+ solver. The basic assumptions for the simulation were as follows: The free surface was neglected in order to reduce the convergence time; The dynamic trim and sinkage of the hulls were also neglected; these simplifications are justified by low or moderate Froude numbers of the vessels (0.16 for the bulk carrier and 0.24 for the car carrier); Only the nominal wake field was computed the propeller operation was not modelled. The unstructured computational mesh of hexahedral cells was used for the computations. The number of cells was varying between 2 and 4 millions of cells, depending on the variant. The visualizations of mesh details are presented in figures 9 11.

Figure 9 Mesh density in the aft part of the hull car carrier

Figure 10 Mesh density in the aft part of the hull bulk carrier

Figure 11 Mesh density in the region of the vortex generators

The solver settings for the simulation were as follows: Turbulence model: Reynolds Stress Transport (RSM); Time step: 0.005 s; Number of iterations per time step: 5. For each case, the flow was considered converged when the averaged value of axial velocity in the propeller disc.

4. RESULTS 4.1 Results of computations for bare hulls and comparison with the experiment. In order to verify the reliability of the CFD simulations, the flow for bare hull was computed for both vessel, and the results were compared with the existing experimental results. The comparison is presented in figure 12.

Car carrier

Bulk carrier

Experiment

CFD

Experiment

CFD

Figure 12 Comparison of the CFD results for bare hull with the experimental results

The following conclusions can be done, basing on the comparison presented above: For car carrier hull, the agreement of CFD results with the experiment can be considered good; For bulk carrier, the area of low axial velocity close to the centre of the propeller disc was underestimated in the computations; however, the overall agreement can be considered sufficient and it can be stated that the numerical model used for the investigation is accurate enough.

4.2 Results of computations for the car carrier appended with vortex generators The results of the computations for hulls with vortex generators were compared with the results for bare hull. For each variant, the comparison includes: Visualization of the velocity field in the propeller disc hull with generators vs. bare hull; Averaged value of axial velocity in the propeller disc, in relation to the average axial velocity for bare hull; Relative value of hull resistance (the free surface was neglected, so the computed resistance does not include the wave resistance; the comparison is then estimative only).

Table 8 Summary of the results for car carrier with vortex generators

Configuration Wake flow (vs.bare hull) No. 1 Bare hull With genetators

Relative average Resistance axial velocity (relative to [%] bare hull) [%] 107.1 107.1

109.7

102.5

102.1

100.9

Summary of the results for the carrier yields the following conclusions: For each configuration, application of the vortex generators resulted in the increase of the average axial velocity in the propeller disc. For configurations 1 and 2, this increase is considerable; However, for each configuration, the resistance is larger than for bare hull; Second configuration of vortex generators is the best compromise between the wake field improvement and increase of resistance.

4.3 Results of computations for the bulk carrier appended with vortex generators The results for bulk carrier are presented in Table 9 in the same way as for the car carrier.
Table 9 Summary of the results for car carrier with vortex generators

Configuration Wake flow (vs.bare hull) No. 1 Bare hull With genetators

Relative average Resistance axial velocity (relative to [%] bare hull) [%] 109.4 88.8

106.3

91.3

For the bulk carrier, the results can be summarized as follows: For both cases, not only the wake flow improvement was achieved, but also the resistance was reduced; Better results were achieved for first configuration of the vortex generators.

Reduction of the resistance by applying the vortex generators can be explained basing on the comparison of the streamlines in the aft part of the hull Figure 13

Figure 13 Bulk carrier streamlines for bare hull and hull with generators

In figure 13, it can be clearly observed that applying the vortex generators results in strong reduction of the vortex intensity and the area of flow separation. The following figure shows the influence of the vortex generators on the axial velocity field. The contour maps of axial velocity are presented in selected transversal planes, for bare hull and for the hull with vortex generators (configuration 1). Influence of the vortex generators can be easily observed in the region marked with the circle, which was filled with the flow of relatively high velocity.

Figure 14 Axial velocity field in the aft part of the bulk carrier bare hull vs. hull with generators

5. CONCLUSIONS The investigation of the influence of vortex generators on the ship wake flow was carried out basing on two test cases: car carrier and bulk carrier. Reliability of the computations was verified on the basis of experimental results. Improvement of the wake flow was obtained for both vessels, however, for car carrier it was obtained at the cost of considerable increase of resistance. For the bulk carrier, very promising results were obtained: the average velocity and uniformity of the wake flow were considerably improved, and the computed resistance of the hull with vortex generators is lower than the resistance of bare hull. Further work on this topic will include: Experimental validation of the CFD results for the bulk carrier hull with vortex generators; Further optimization of the configuration of vortex generators; Attempt on elaborating the guidelines for setting up the configuration of vortex generators for improving the wake flow of large vessels.

ACKNOWLEDGEMENTS The research presented in this paper has been financially supported by The Polish National Centre for Research and Development (NCBiR), Grant No. N R10 0040 06/2009. The authors would like to express their gratitude for this support.

REFERENCES Schmde D., (2008), RANS computations for wake improving vortex generators, Proceedings of 11th Numerical Towing Tank Symposium (NuTTS), Brest, France.

CONTACT DETAILS Pawe Dymarski, Centrum Techniki Okrtowej S.A., ul. Szczeciska 65, Gdask, Poland pawel.dymarski@cto.gda.pl Marek Kraskowski, Centrum Techniki Okrtowej S.A., ul. Szczeciska 65, Gdask, Poland marek.kraskowski@cto.gda.pl

Pre- and Postswirl Devices in View of their Optimum Adjustment to the Propeller Layout
Heinrich Streckwall Hamburg Ship Model Basin, Hamburg, Germany

The recovery of the rotational energy which is contained in the propeller slipstream is addressed. The quality of pre- and post-swirl devices installed for this purpose upstream or downstream of the propeller is usually judged from model tests. This procedure can be labelled a try and error approach and is not really reflecting a sound process that guarantees a finally optimized setup. Therefore a strong need for an improved practice is identified and accordingly the current paper proposes to support experiments by a computation phase to find global adjustments in advance. We assume traditional pre-and post swirl systems consisting of vanes mounted co-axial to the propeller axis and want to find the optimum geometry. According to a paper of Betz (1938) the computational optimum is related to a prescribed vane-circulation. As specific propeller characteristics cannot be ignored we classify the propeller first and adjust the vane parameters subsequent. For the vanes we apply a panel method, which allows us to study parameters like the number of vanes arranged on the pre/post swirl unit and the radial extension of these vanes. It is the aim of the paper to reflect the inter-dependency of wane design and propeller design.

Keywords: stator, propeller, spiral energy, energy recovery, contra-vanes, swirl flow.

1. INTRODUCTION It can be notified, that the efforts to improve propeller propulsion using pre- or post-swirl devices has increased recently. Figure 1 shows the general flow conditions for a rudder (as a representative of a post-swirl device). These devices are mainly installed to arrive at a propulsor (the latter consisting of propeller and stator) with a slipstream that shows strongly reduced circumferential components. The quality of pre- and post-swirl devices is usually judged from model tests. As it is the aim to reduce power consumption, the decisive experiment is the self propulsion test in the towing tank. During several runs the installed vanes that build up the pre-/post swirl unit are turned stepwise around their axis to achieve a new configuration subject to new test runs. Keeping the speed of the ship model constant, the set of runs is finally analysed with respect to the least power measured at the propeller. This try and error approach may lead to a long process to achieve a finally optimized setup. The scope of such tests hardly includes measurements of the propulsor slipstream characteristics, which could be a way to typify the propeller and approximately adjust the vanes beforehand. We identified a strong need for an improved practice and in the following present some basic results from a computational tool. This numerical instrument is supposed to identify potential efficiency gains and serves to specify the required geometry parameters to achieve such gains in reality.

Figure 1: Sketch for general inflow conditions to a rudder (formally representing a system of 2 contra-vanes). Solid curve for a typical axial velocity profile, dashed for a typical tangential component (changing the direction of the angle of attack at the propeller shaft height).

2. THE SCOPE OF THE STUDY PERFORMED BY BETZ In his paper, Betz (1938) gives very general recommendations for an optimal propulsion system consisting of a ship propeller (rotor) and a stator mounted for the sake of energy

recovery in front or behind the propeller. He uses the terms spiral energy and contra-vanes expressing that the re-direction of the swirl flow is the objective of the system. As he considers most of the spiral energy left behind the propeller as recoverable, he assumes a propeller design, which is optimized in a sense that it shows (as far as possible) a constant axial induced velocity component up all across the slipstream. The tangential velocity profile of the slipstream related to such a propeller design is quite simple, as its shows exactly a 1/r decay from the hub to the outer border or the slipstream. For these boundary conditions he studies contra-vane systems categorized as follows: 1) Vanes with infinite radial extent distinguished by: i) ii) Different blade number ZV Different hub diameter d i (more generally different ratios i of hub radius ri to the outer slipstream radius R ; where generally denotes r / R )

2) Vanes reaching up to the slipstream boundary or staying inside the slipstream

distinguished again by ZV and i . 3) Vanes of finite radius crossing the slipstream boundary distinguished once more by ZV and i . For each of these configurations he gives the optimum circulation distribution along the vane radius and (neglecting friction at first) also the recovery-efficiency, simply named . The vane circulation ( ) is however already a derived quantity. An even more basic demand is the cancellation of the propeller swirl flow achieved by the downwash of the contra-vanes (downwash taken far behind the individual vane). This guarantees vanes operating neither with excessive circulation (creating too much downwash and thus an unfavourable force direction) nor with too moderate circulation (creating an unfavourable low force magnitude). For the latter quantity he introduces the infinitely bladed stator, which achieves the total recovery of spiral energy as long as it reaches the slipstream boundary. Any device of finite blade number and finite extent also achieves a complete recovery if the hub diameter tends to zero! This is however a hypothetic limit, as in parallel the spiral energy content of the slipstream reaches infinity. According to Betz it is favourable to use more than two vanes for reasonable hub to slipstream ratios ( ri 0.2 ). It also improves the performance if the vanes pass the slipstream boundary.

3. THE SCOPE OF THE CURRENT STUDY 3.1 Relating Vane Circulation and Vane Geometry for Post Swirl Position
Extending the work of Betz we will come up with actual stator geometries for post swirl application. The geometries succeed to establish the desired circulation distribution. The process of creating such vanes is an iterative one, but as the vane geometry parameters, namely local pitch and local camber are quite linearly linked to the local circulation, it is a deterministic process leading quite straight forward to the optimum. As the panel method determines (by pressure integration) axial forces acting on the stator, it can be checked if their maximum contribution to the thrust coincides with a reproduction of the target circulation. Within a panel method, the circulation can be taken from the jump of the potential at the trailing edge. For arbitrary propeller characteristics it can become quite involved to relate the optimum circulation on the contra-vanes ( ) to the swirl flow of the slipstream, the latter reflecting an equivalent total propeller circulation o ( ) = Z p () . Therefore we were looking for an alternative indicator.

3.2 Complete Recovery in Post Swirl Position

One may define different inflow profiles vt (r ) which remain unchanged along the circumference. Betz (1938) assumed a vt (r ) -profile related to a (Z bladed) propeller with constant total circulation Z p (r ) = = const , which reads:

0; r < ri vt ( r ) = v t ( r ) = /(2 r ) ; ri r R 0; r>R

(1)

where ri is the hub radius and R is the slipstream radius. Following Betz (1938) we establish the condition for the optimum self induced downwash w(r ) at the lifting line of the contra-

vane:

1 w(r ) = vt (r ) 2

ri r

(2)

This is the condition for the optimum flow at a contra-wane. It implies that the induction in the wake of each vane = 2 w(r ) cancels locally the incoming swirl. Betz (1938) derives the vane circulation (r ) related to this optimum. An interpretation of his findings is, that with finite Z v one will not succeed to annul the incoming swirl in the circumferential mean. This also implies that the total vane circulation Z v (r ) will not reach the circulation which drives the incoming swirl, namely Z p (r ) and simultaneously fulfil the optimum condition Equ. (2). Coincidence of this needs the infinite bladed stator ( Z v = ). However the stator with Z v = may serve as a reference to judge the quality of thrust recovery. A reference thrust
S o related to the total annulment of the swirl reads:

S o = 1 / 2vt (r ) Z p (r )dr
ri

(3)

Any actual thrust S related to an actual total vane circulation Z p (r ) will be lower and thus (following Betz) one can define a stator efficiency as:

S S o

(4)

In case the constant total propeller circulation Z p (r ) = drives the swirl, i.e. in case of a

vt (r ) as given in Equ. (1) the reference thrust S o reads: S o = 1 / 2 R ( ) 1 /(2 r ) dr = 41 ( ) ln r

2 R 2 ri i

(5)

3.3 The Link to Global Propeller Parameters

On the other hand, neglecting the self induced tangential component at the lifting line, a propeller with constant total circulation Z p (r ) = would deliver the thrust:

S = r dr = ( / 2) ( R 2 ri )
2
ri

(6)

so that in this case:

S o 1 (ln R ln ri ) = S 2 R 2 (1 ri 2 / R 2 )

(7)

From Equ. (6) we may relate the thrust coefficient K T = S /( n 2 D 4 ) and giving
2 (1 ri / R 2 ) = 8 K T / 2 R 2 so that (8)

(ln R ln ri ) S o = (4 K T / 3 ) 2 S (1 ri / R 2 ) 2

for ri / R = 0.2 :

S o = 0.225 K T S

(9)

Also for propellers showing a non-constant circulation over the radius, the right part of the above equation may serve as a rough estimate to relate the maximum relative thrust recovery to the propeller thrust coefficient. Finally we have to establish a reasonable axial slipstream component. A normalized total axial velocity u + 2u is obtained via ( CTH = K T / J 2 8 / ):

u + 2u = 1 + CTH

(10)

3.4 Pre-Swirl Stators

In this paper, pre-swirl stators will not be treated explicitly. Their hydrodynamic design can be considered nearly accomplished, if a post-swirl stator has been found which compensates the spiral flow behind the propeller appropriately. They just have to reproduce the distribution of circulation (r ) which was found optimal for the post-swirl situation. Compared to the post-swirl case they need an adjustment for the pitch so that they operate at least at the same angle of attack in front of the propeller as they would experience behind the propeller due to the spiral flow lines. To duplicate the target circulation (r ) some further increase of pitching is necessary to compensate for the reduced absolute flow speed in front of the propeller.

Figure 2: A two-bladed post swirl stator

4. COMPUTATIONAL RESULTS 4.1 Stator in an Ideal 1/r-Type Inflow

Using a Boundary Element Method - similar to the approach Fine and S.A. Kinnas (1993) applied to the propeller - for the stator we considered a case, where a 2-bladed stator was subject to an idealized propeller slipstream. The stator (see Figure 2) showed no hub and a radial extension Ro of 1.5 times the slipstream radius ( Ro / R =1.5). The idealized slipstream was described according to the following procedure: a) We fixed a propeller thrust coefficient K T =0.193 and assumed a hub diameter to blade diameter ratio of 0.2, in other words ri / R =0.2. b) We fixed the J-value (J= u /(nD ) ) to 0.814 and obtained CTH = 0.741 and derived from Euq. (10) an axial slipstream component u + 2 u = 1.32. c) We deduced the total circulation of an equivalent propeller with constant circulation using Equ. (8) and obtained the tangential component at the stator vt (r ) via Equ. (1). We extended the profile down to r = 0 keeping the value vt ( ri ) inside r ri .

Figure 3: Normalized tangential inflow with 1/r profile

1.00 0.80

Panel
0.60 0.40 0.20 0.00 0 0.25 0.5 0.75 1 1.25 1.5

Betz

Figure 4: Comparing solutions for ciculation based on the total propeller circulation o . The panel solution is related to the maximum stator thrust and was found iteratively.

As can be normalized as

, the component vt (r ) resulting from Equ. (1) can for R 2

instance be normalized as vt ( r ) /(R ) , which is plotted for the specific case K T =0.193 in Figure 3. The geometry of the stator vanes was adjusted to achieve the maximum thrust S . The stator thrust S was obtained via surface pressure integration. For the stator vanes the basic geometry parameters influencing the pressure distribution are chord, camber and pitch. Our adjustments were done solely on pitch and camber. It was registered, that the proposals on the optimum circulation distribution given by Betz (1938) served as a good target circulation to guide a try and error approach (which was still the way how we adjusted the geometry) quite straight forward to the thrust maximum. Figure 4 compares with the Betz solution for the total circulation on a 2-bladed stator subject to the flow of Figure 3, characterized by ri / R =0.2 (his solution actually assumes the vanes to be mounted on an infinite hub and Ro / R =1.5. Normalizing the stator thrust S (obtained from the pressure integration, neglecting friction) in the same way as the thrust S is normalized in K T and building the quotient we arrived at S / S =0.029, meaning roughly 3% added thrust due to thrust recovery. Table 1 summarizes the whole test case, which is basically defined by the tangential component of Equ. (1).
Table 1: Settings / results for test case defined by the tangential comp. given in Equ. (1). Ideal propeller Stator

KT
0.193

J 0.814

u + 2u /(R ) Z v

ri / R
0.2

Ro / R
1.5

S o / S
0.0434

S / S
0.029

1.32

0.163

4.2 Results Related to a Calculated Slipstream

Next we design a propeller which shows a (nearly) constant circulation over the radius. The panel system related to such a design is shown in Figure 5. Having set up such a geometry an unsteady propeller panel method (in house code PPB) was applied to evaluate the propeller induced velocities at the rudder axis. In this context PPB serves mainly to specify the rudder inflow and delivers the propeller induced velocity components registered along the rudder axis. Practically we position sensor points along an imaginary rudder shaft and determine mean values for all velocity components as an average of the components that relate to

specific angular positions of the reference blade. For simplicity propellers were considered to operate in homogeneous inflow without initial swirl.

Figure 5: Discretization of propeller blade representing the panel system related to a physical solution (here Z=8) for the Betz-propeller, accordingly showing a constant circulation over the radius (and in the limit of an infinite number of blades creating a slipstream with homogeneous axial velocity profile).

v_t

0.6 0.5 0.4 0.3

tangential in Slipstream

0.2
Tang

0.1 0 0

< 1/r character.

r/R

0.2

0.4

0.6

0.8

Figure 6: Tangential velocity profile of propeller shown in Figure 5 together with an exact 1/r-swirl

Now the character of the tangential slipstream is close to (but not identical with) the ideal 1/r behaviour (Figure 6). Nevertheless the actually achieved circulation distribution on a 2 bladed vane system extending to 1.5 slipstream diameters is again very close to Betz optimum (Figure 7). The efficiency =
S obtained with the panel method even slightly increased S o

(Table 2). However in this case the thrust coefficient (0.214) was higher than before (0.193).

1.00 0.80

Panel
0.60 0.40 0.20 0.00 0 0.25 0.5 0.75 1 1.25 1.5

Betz

Figure 7: Circulation distribution on 2 bladed vane system with Figure 6; for the panel calculations the reference G o

Ro / R =1.5 subject to tangential flow shown in

= is estimated according to Equ. (8); Betz optimum

for same system reproduced from Figure 4. Table 2: Settings and results for the test case characterized by the tangential component given in Figure 6 Real propeller Stator

KT
0.214

J 0.814

u + 2 u v t
Figure 6 Figure 6

Zv
2

ri / R
0.2

Ro / R
1.5

S o / S
0.048234

S / S
0.033

4.3 Alternatives to Specify the Optimum Circulation

Betz (1938) used special transformations to derive the optimum circulation for finite values of Ro / R analytically. However his inflow conditions vt (r ) were confined to the one vt (r ) function given in Equ. (1). In the beginning of his paper he also gives some hints, that vanes of infinite axial and radial extent would represent perfect solutions for the task to compensate an incoming swirl flow, if they would not suffer from an infinite drag. This gave rise to the idea to establish panel systems for rectangular wings of different axial extent. They are shown in Figure 8, representing very simple solutions for a 2-bladed stator.

Figure 8: Rectangular 2 bladed stators of various aspect ratios w/o pitch and camber, chord c/R=2, 1 and 0.5

When the stators given in Figure 8 were exposed to the tangential inflow shown in Figure 3 we obtained the normalized circulation distributions of Figure 9. It is impressive how

regularly the computed circulation comes closer to the optimum given by Betz when the axial dimension is increased.
G/Go

1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 0

Various aspect ratios

Betz Panel 3 Panel 2 Panel 1

r/R

0.25

0.5

0.75

1.25

1.5

Figure 9: For a stator with 2 rectangular vanes the extension of chord (1, 2 and 3 correspond to the stepwise increase of chord given in Figure 8) seams to lead to the optimum circulation

Figure 10 : Rectangular 4 bladed stators (including vortex wake) w/o pitch and camber, chord c/R=2, 1 and 0.4

4.4 Changing Number of Vanes and their Radial Extent

As we did for the 2-blade stator with Ro / R =1.5 we also investigated 4-bladed stators of rectangular vane shape, all showing Ro / R =1.5, zero camber, zero pitch but different chord. There geometry is displayed in Figure 10. Computed results in terms of the normalized circulation are displayed in Figure 11 together with the optimum 4-bladed solution derived by Betz. Already for the medium stator there is a striking similarity to the Betz result.

G/Go

1.00 0.80 0.60 0.40 0.20

4 Blade Stator Series Betz Panel 1 Panel 2 Panel 3

r/R 0.00 0 0.25 0.5 0.75 1 1.25 1.5

Figure 11: For a stator with 4 rectangular vanes the extension of chord (1, 2 and 3 correspond to the stepwise increase of chord given in Figure 10) seams to lead to the optimum circulation. The solution of Betz is now also accounting for 4 blades.

For a 2-bladed vane system the performance loss due to a reduction of the vane radius Ro appears to be moderate as shown in Figure 12. To interpret the difference in circulation one may refer to the Kutta-Joukowski lift theorem and evaluate the axial force f x acting between
r and r + dr according to f x= v ( r ) dr , assuming further that the resultant normal

inflow to the sections v can be sufficiently approximated by vt / 2 inside R and by 0 outside.

1.0

0.8 Panel 2 0.6 Panel 2 Ro/R=1

0.4

0.2

0.0 0 0.25 0.5 0.75 1 1.25 1.5

Figure 12: Loss in performance for 2-blade stator due to a reduction from

Ro / R

=1.5 (Panel 2) to

Ro / R

=1.0 (Panel 2, Ro/R=1)

5. SUMMARY AND CONCLUSION

An HSVA internal surface panel method RPB, similar to the internal propeller panel method PPB (both RPB and PPB can also be categorized as BEM =Boundary Element Method) was applied for the stator analysis. Here we used an extreme propeller design, characterized by a perfect 1/r- behaviour of the tangential slipstream component. A material propeller has been

designed to figure out, whether a physical object is able to create the ideal 1/r shape. This material propeller with 8 blades did not significantly diminish the stator performance. Ideally, if friction could be ignored, it represents the optimal design for a propeller/stator combination. Though in principle possible via simple drag coefficients, we did not investigate quantitatively how drag effects would influence our conclusions. The two basic findings however where: 1) We could reproduce the results given by Betz (1938) in terms of an optimal vane circulation by maximizing the vane thrust, the latter obtained by a surface pressure integration. 2) We propose a practical way to derive the optimum vane circulation, namely analysing an artificial vane with extreme chord. The procedure was actually applied and tested against the circulation given by Betz (1938) for the 1/r swirl flow. It seems reasonable to proceed in the same manner for any propeller slipstream characteristic. Also when dropping the constraint of circumferentially constant velocity fields the analysis of the artificial vane may help to optimise stators that are supposed to operate in a velocity field resulting from a ships wake and added propeller induction. This was actually the motivation of the current study as the pure try and error procedure of repeated towing tank tests seems to need some support.

5. REFERENCES
Betz, A.(1938): Zur Theorie der Leitapparate fr Propeller. Ingenieur-Archiv, Vol. 9, No. 6, December 1938 (also available under The theory of contra-vanes applied to the propeller in
a NACA Technical Memorandum, No. 909, Sept. 1939, pp. 435-452)

Neal E. Fine and S.A. Kinnas (1993): A Boundary Element Method for the Analysis of the Flow Around 3-D Cavitating Hydrofoils, Journal of Ship Research, Vol. 37, No. 3, 1993. 3-D Cavitating Hydrofoils, Journal of Ship Research, Vol. 37, No. 3, 1993.

CONTACT DETAILS
Heinrich Streckwall, HSVA Hamburg Ship Model Basin, Bramfelder Strae 164, Hamburg streckwall@hsva.de

Experimental and numerical modelling of cavitation and erosion on rudder

Pawe Dymarski Centrum Techniki Okrtowej S.A.

This paper presents results of experimental tests and numerical simulation of cavitation on rudder of a fast twin-propeller ship. The paper includes short description of mathematical and numerical bases of the computational model of flow with cavitation. Two types of rudder have been taken into consideration: conventional (symmetrical) rudder and flow adapted twisted rudder. The results from CFD analyses are compared with results obtained from model tests with the use of High Speed Video camera. Additionally, the results of soft paint tests of erosion on rudders are presented.

Keywords: rudder, cavitation, erosion, propeller, RANSE

1. INTRODUCTION The main target of the project, which results are presented in the article, was to develop computational method for modelling of cavitation phenomenon on rudder behind a propeller. The model was developed in order to enable prediction of cavitation and the secondary effect of the phenomenon erosion on the surface of rudder. The second part of the computational model prediction of erosion is not finished yet, however the first part - visualization of cavitationon on rudder has been already tested and the preliminary results of this part of the project are presented below.

2. MATHEMATICAL MODEL 2.1. GOVERNING EQUATIONS OF THE RANSE MODEL The closed system of motion equations, derived for incompressible fluid, is based on the momentum and mass conservation laws. An integral form of mass conservation equation formulated for control volume with a surface S reads

v ndS
S

(1)

and the conservation equation of i-th momentum component has the following form:

u i d + u i v ndS = ( ij i j pii ) ndS , t S S

(2)

where v is velocity vector, ui - i-th velocity component, p - pressure, - density, n - unit vector normal to S surface, ii - i-th component of Cartesian unit vector and ij is a viscous stress tensor. When the flow is turbulent, v and ui refer to mean velocity vector and mean i-th velocity component, p is a mean value of pressure. The word "mean" denotes average in a time period, which is long compared to the period of turbulent oscillations (Wilcox, 2002). The viscous stress tensor ij is specified by Boussinesq approximation presented by Blazek (2001) and Wilcox (2002):
ij = 2( + t )S ij , (3)

where is a molecular viscosity, t is the turbulent viscosity and S ij is the mean strain-rate tensor. The turbulent viscosity is calculated with the use of Spalart-Allmaras turbulence model presented by Blazek (2001) and Wilcox (2002).

2.2. NUMERICAL MODEL OF PROPELLER

Within the introduced model the screw propeller is being modelled in the simplified way, using nonstationary function of volume forces. In order to define the distribution of volume forces one should implement the geometry of the contour of the propeller's blade and the field of axial and circumferential forces on the blade. The implemented field of volume forces is being normalised in order to get specified total values of the propeller thrust and the torque. Defined force field is rotating with the determined angular speed around the propeller's axis, inducing the unsteady, periodic field of velocity. The pressure distribution on the blade, which is needed for modelling of the propeller, can be evaluated from codes based on potential methods (for example solvers based on Boundary Element Method), as well as it can be approximated using simplified functions of pressure distribution. The latter method was used for computations of test cases presented in this paper. The 2D function of blade forces was build using two functions. The first one describes the forces distribution along blade section, the second function describes a value of maximum of the first function along the radius of propeller. Both of them are normalized, therefore the maximum value is one for both. Therefore the functions of volume forces in local co-ordinate system is as follow: f T = aT
max

f sec (u ) f rad (r ) f x ( x ) , f sec (u ) f rad (r ) f x ( x ) . (4)

f Q = aQ

max

Where: fT , f Q f sec u f rad r fx aT

max

axial and circumferential component of volume forces; function of volume force distribution along blade section; nondimentional coordinate, which defines position along chord of section: u = 0 at leading edge; u = 1 at trailing edge; distribution of MAX ( f sec ) along blade span; nondimentional radius, which defines position along blade span: r = 1 at the tip of the blade; function, which determines the thickness of acceleration disk;

, aQ

max

is a maximum value of axial and circumferential component of volume force.

2.3. MODEL OF CAVITATION

The cavitation model is based on travelling bubble method presented by Lecoffre (1999) and Dymarski (2007). It is assumed in the model, that a large number of micro gas nuclei is present in the liquid. When pressure value decreases below a specified critical level, the radius of nucleus starts to grow rapidly and according to the model this is the inception of cavitation. To determine behaviour of a single bubble the pressure field, velocity field (or bubble trajectory) and initial size of nucleus have to be given. The single bubble dynamic is described by Rayleigh-Plasset equation (Lecoffre, 1999):

d 2 R 3 dR dR R 2 + + = dt R dt 2 dt

p+

2A p p g R ,

(5)

where: R is a radius of the bubble, t is time, p is pressure of water affected on the bubble, pv - vapour pressure, p g pressure of the gas in the bubble, A is the surface tension. The value of p is evaluated as a mean value of pressure at the surface of bubble:

pmean =
where S is a surface of the bubble.

1 p(x, y, z )dS , S S

(6)

The pressure of gas inside bubble is calculated with the assumption that the process of growing and collapsing is isothermal:

R pg = pg 0 , 0 R
Where p g and R0 is pressure of gas inside bubble and radius of the bubble at inlet.
0

(7)

The introduced model of predicting the phenomenon of cavitation is a hybrid model. The velocity field and the pressure field are being calculated with finite volume method using the grid defined in main system of coordinates, however dynamics of bubbles are being solved in local frames of reference moving with bubbles. In order to obtain the information about the volume of the gas phase inside control elements of the computational mesh, an additional, fine grid is constructed. The new auxiliary grid is structural and orthogonal with cubic elements. Each of them is storing the information about the volume of gas phase and the

index of the element of the main grid in which it is included. The algorithm for calculating of the gas phase volume is as follows:

1. 2. 3.

For every bubble elements of the auxiliary grid which are inside it are being found Calculating volume of the gas phase included in elements of the auxiliary grid Calculating the volume of the gas phase in elements of the main grid, by adding the

volume of gas from elements of the auxiliary grid.

Because capacity of elements of the auxiliary grid is limited, the algorithm prevents multiple summing up the gas volume from overlapping bubbles. When a volume of gas in a control volume of the main grid is available, then the volume fraction of gas phase inside specified CV can be evaluated: = Vg VCV , (8)

Where, VCV is volume of control element, = 0 - when the volume VCV is fully filled with water, = 1 - when the volume VCV is filled with gas.

With a view to accelerate the computational process, as well as in order to reduce the number of the required RAM, the phenomenon of the cavitation is being estimated only in the defined relatively small area inside the domain of flow computation. In case of calculations of the cavitation phenomenon on a rudder model, computational area is included inside the cuboid with a breadth about 4 cm bigger than the thickness of the rudder (in the model scale) and of the height and the length of the subdomain about 20 - 50 % greater than the relevant size of the rudder. The narrow zone at the inlet into the cuboid, is an area, in which microbubbles of gas are injected to water. Initial position of microbubbles is determined through random process, the amount and sizes of bubbles are being selected on the basis of statistical data.

3. DESCRIPTION OF THE TEST CASES

The test cases are based on condition of real case of fast twin propeller boat, which was tested in CTO's cavitation tunnel a few years ago. The presented results correspond to half load conditions of the vessel.

In order to extend the analysis the additional test case was prepared, a rudder with alternative geometry has been designed based on preliminary CFD results with conventional rudder (Figure 1). The shape of the new rudder was adapted to geometry of flow behind a propeller, the peculiar shape of leading edge causes, that the pressure difference is smaller compare to the original shape, and hence the risk of the appearance of cavitation (or the scale of the phenomenon) is smaller. The experiment with the new shape will allow for better testing the developed computational model. If the differences between the test cases in scale of cavitation phenomenon obtained from the numerical model are similar to ones observed in cavitation tunnel, the model can be useful in design process.

Test conditions (in model scale):

pressure at propeller axis: p0 = 10200 Pa ; velocity at inlet: V = 3.83 m/s; propeller geometry: D =0.250 m ; P0,7/D =1.174; Z = 3;
1 propeller revolutions: n p = 16 ; s thrust and torque coefficient: K T = 0.136; K Q = 0.056 ;

Figure 1. Comparing of the geometry of models of two rudders, for which experimental tests and computational analyses were performed as part of the project. On the right-hand side: rudder with conventional geometry, on the left: rudder with the leading edge twisted, adapted to the geometry of the velocity field.

3.1. FLOW SIMULATION WITH CFD

The CFD domain was a copy of geometry of the dummy-body model in cavitation tunnel. The computational grid was block-structural and heksahedral. The grid was generated with the use of ANSYS-ICEM HEXA. The grid structure shows Figure 3.

Figure 2. The structure of grid for computations of flow around conventional rudder. Number of control volumes was about 3.7 M.

The results of computations of cavitation phenomenon on conventional rudder are shown in Figure 3. The results for twisted rudder are presented in Figure 4.

a)

b)

Figure 3. Results of computations of cavitation on conventional rudder (behind a propeller). The time interval between successive frames is (in the model scale) 0.004 s , a) portside, b) starboard view.

a)

Figure 4. Results of computations of cavitation on twisted rudder (behind a propeller), a) starboard view. (No cavitation has been obtained due to computation at portside).

3.2. MODEL TESTS RESULTS

The model configuration in test section of cavitation tunnel is shown in Figure 5 a). The results of model tests for conventional rudder are presented in Figures 6, 7, however tests results for twisted rudder are shown in Figures 8, 9.

a)
of the model

b)

Figure 5. Configuration of after-body model in the CTO's cavitation tunnel: a) draft of the test stand, b) picture

Figure 6. Model tests results: cavitation on the conventional rudder. Snapshots from High Speed Video recording. The time interval between frames is 0.002 s. View from the port side.

Figure 7. Model tests results: cavitation on the conventional rudder. Snapshots from High Speed Video recording. The time interval between frames is 0.002 s. View from the starboard.

Figure 8. Model tests results: cavitation on the twisted rudder. Snapshots from High Speed Video recording. The time interval between frames is 0.002 s.View from the port side.

Figure 9. Model tests results: cavitation on the twisted rudder. Snapshots from High Speed Video recording. The time interval between frames is 0.002 s. View from the starboard.

3.3 MODELLING OF CAVITATION EROSION SOFT FILM TESTS

In order to predict an area peculiarly exposed to erosion a special type of test has been performed. The same method which is applied for prediction of erosion on propeller blade was applied to predict the phenomenon on the rudder. The model of the rudder was painted with special soft paint to assess the presence of cavitation impingement. After 120 minutes in specified constant conditions in cavitation tunnel the rudder was subjected to detailed inspection. The defect of soft coat on the surface of model of conventional rudder has been noticed (Figure 10 a). In order to make sure that the received result isn't accidental, the erosion test has been repeated. The results from the second test were almost the same. Twisted rudder has been tested according to the same procedure, but no defect of the coat was identified (Figure 10 b).

a)

b)

Figure 10. Models of rudders after soft film tests of cavitation erosion: a) the coat of conventional rudder has been defected, the arrow indicates places, in which bare material (bronze) is visible b) the covering of twisted rudder stayed intact.

4. CONCLUSIONS AND REMARKS:

the presented computational model seems to be sufficient to predict more significant

structures of cavitation on rudder,

the difference between cavitation on conventional and twisted rudder has been captured

by the model, however some forms of cavitation were underestimated (or just omitted),
the evolution of main cavitation structures due to non-steady periodic flow obtained from

computation was very similar to the cavitation evolution recorded by HSV camera,
some structures of cavitation were strongly underestimated, for example the cavitation

behind a sharp leading edge near the bottom of the rudder in Figure 6 and Figure 4 a),
some weak structures of cavitation were not predicted by the model, for example

cavitation at the starboard side on the twisted rudder Figure 9,

the soft paint test, which is usually applied for prediction of cavitation erosion on

propeller's blade seems to be applicable for investigation of erosion on rudder Figure 10,

ACKNOWLEDGEMENT
The research presented in this paper has been financially supported by the Polish Ministry of Science and Higher Education Grant No. N504 489134. The author would like to express his gratitude for this support.

REFERENCES

Blazek J. (2001), Computational Fluid Dynamics: Principles and Applications, ELSEVIER Dymarski P. (2007), Predictions of the open water propeller cavitation using the SOLAGA solver, 9th Numerical Towing Tank Symposium, 23-25 September 2007, Hamburg-Germany Lecoffre Y. (1999), Cavitation. Bubble Trackers, A.A. Balkema/Rotterdam/Brookfield Wilcox D.C. (2002), Turbulence Modeling for CFD, DCW Industries

CONTACT DETAILS
Pawe Dymarski, Centrum Techniki Okrtowej S.A., ul. Szczeciska 65,Gdask, Poland; pawel.dymarski@cto.gda.pl

Effect of propulsion system parameters on fuel consumption by inland vessel

Jan Kulczyk, Emilia Skupie Hydraulic Machines and Systems and Inland Vessels Modelling Division of Wroclaw University of Technology, Wrocaw, Poland

Fuel consumption has a significant effect on the operating costs of inland vessels. It is possible to estimate fuel consumption for the prescribed operating conditions on the basis of the propulsion systems parameters. The method described in this paper is based on a mathematical model of the hydrodynamic interactions in the propulsion system of the inland vessel. The power demand for the given operating conditions is calculated using the vessel and propulsion system parameters as the input data. Then fuel consumption is determined on the basis of the engine characteristics. The method allows one to analyze the effect of individual parameters on the operation of the propulsion system. This paper presents the results of tests of the newly upgraded propulsion system of the Bizon III push tug. The upgrade included the replacement of the main engines and a change of the propeller pitch. The presented analysis concerns the effect of propeller pitch on fuel consumption in given operating conditions. The influence of other propeller parameters is also analyzed. A range of operation in which the main engines are free from overload was determined for the upgraded propulsion system.

Then the operating parameters optimum with regard to fuel consumption were determined. It turned out that the propeller pitch is optimal only in a narrow range of vessel speed but it is adequate to the actual conditions in which the Bizon III pusher tug operates.

Key words: fuel consumption, propulsion system, push tug Bizon III , propeller pitch, mathematical model

1. INTRODUCTION Today fuel costs are the principal component in the total inland waterway transport costs. Fuel consumption is determined by the vessels resistance to motion and the way in which power is transmitted by the propulsion system. One can shape the total resistance only by changing the shape of the vessel, but this would be very expensive. Alternatively, one can reduce fuel consumption by modifying the propulsion system. For this purpose the effect of the particular propulsion system parameters on fuel consumption in the given operating conditions should be examined. This paper presents an analysis of the effect of vessel propulsion system parameters on fuel consumption. The analysis was carried out using a method based on a mathematical model of the hydrodynamic interactions in the propulsion system of the inland vessel.

2. DESCRIPTION OF THE METHOD The HPSDK computing system, developed in Hydraulic Machines and Systems and Inland Vessels Modelling Division of Wroclaw University of Technology, was used to analyze the effect of the propulsion system on fuel consumption. The computing system incorporates a model of the hydrodynamic interactions in the propulsion system of the inland vessel. The model takes into account the mutual interactions between the propeller, the vessel hull and the waterway in specified hydrodynamic conditions. Thanks to this method one can analyze the effect of the individual parameters on the operation of the propulsion system. The wake fraction, the propeller thrust and the required shaft torque can be calculated numerically. By modifying the propeller parameters (propeller series, diameter and pitch) one can assess the suitability of the propeller from the fuel consumption point of view for the given state of loading in the specified hydrotechnical conditions. In addition, knowing the thrust demand curve one can make vessel propulsion forecasts for different operating parameters. A computer program (Kulczyk, Sradomski 2006) was used to determine fuel consumption. Depending on the waterway depth and the barge draught the program selects propulsion system parameters at which fuel consumption is minimum. The required input data are: propulsion forecast, range of vessel speed, propeller parameters and propulsion engine performance (including specific fuel consumption) (2006). If there is no propulsion forecast generated by the HPSDK computing system, the program generates it using the approximate method. The

calculations are based on the prescribed resistance curve, the impact factors and the hydrodynamic characteristic of the screw propellers. The program performs the calculations in the EXCEL environment (spreadsheet).

2.1 Verification of method accuracy The accuracy of the method was determined by comparing the calculation results with the shipowner data for two pusher tugs Bizon III with Wola Henschel and Volvo Penta engines. The specifications of the tested propulsion system are shown in table 1 [1975].
Table 1. Specifications of Bizon III pusher tugs engines and propellers. Wola 05H6Aa power rating power transmission efficiency rated engine rpm gear ratio propeller Ka4-55 in nozzle 19A diameter pitch D=1.26 m P/D=0.778 D=1.26 m P/D=0.829 PB=139 kW TR=0.91 ns=1500 min-1 i=4 Volvo D9-300 PB=221 kW TR=0.95 ns=1800 min-1 i=4

Fuel consumption-vessel speed relations for the actual propulsion system specifications were calculated using a spreadsheet taking into account impact factors and hydrodynamic characteristics. Push train draught T=1.6 m at waterway depth h=4 m was assumed. The thrust curve for the prescribed conditions was determined on the basis of model studies (1965). The following impact factors were assumed: wake fraction w = 0.3 and thrust deduction factor t=0.2. The ODRATRANS company supplied data which made it possible to determine fuel consumption as a function of vessel speed for similar hydrotechnical conditions. Fuel consumption was also determined by the conventional method. Propulsion characteristics in the form of the family of curves: T=f1(Vs, Q=const) T=f2(Vs, n=const) and: PD=g1(VS, Q=const) PD=g2(VS, n=const) (3) (4) (1) (2)

were determined from the hydrodynamic propeller characteristics. The propulsion characteristics were the basis for propulsion forecasts PD=f(VS) and n=f(VS). The power demand-rpm curves and the fuel consumption-rpm curves were used to determine fuel consumption by the particular engines in the given navigation conditions (1986) (1965). The results are compared in diagrams shown in fig. 1 (for the Wola Henschel engine) and fig. 2 (for the Volvo Penta engine).

Figure 1. Comparison of fuel consumption versus vessel speed for Wola Henschel engine.

Figure 2. Comparison of fuel consumption versus vessel speed for Volvo Penta engine.

The differences between the theoretical values may be due to the different accuracy with which dimensionless thrust KT and torque KO (as a function of advance J) coefficients and thrust deduction factor t were assumed and readings from the resistance (R) curve and the specific fuel consumption (ge) curve were taken. The interpolation method in the spreadsheet was different than the one used in the direct calculations. The empirical results are biased because of the way in which fuel consumption was calculated (since the pusher tugs with the Wola Henschel engine have no onboard fuel consumption measuring computer, the fuel consumption was estimated) and also due to the fact that waterway conditions are averaged and the specified speed range only slightly coincides with the theoretical values. Moreover the values given by the shipowners are lower than the theoretical ones since experienced ship operators set such engine operating conditions which ensure the lowest fuel consumption.

3. UPGRADE OF BIZON III PUSHER PROPULSION SYSTEM The parameters of the propulsion system of the upgraded Bizon III pusher tug were analyzed. The results of this analysis are presented below. The upgrade consisted in the replacement (because of the exhaust gas cleanliness requirements) of the main propulsion engines. Since the new engines horsepower was higher it was necessary to recalculate and match the other parts of the power transmission system. Ultimately the propeller pitch was changed while the propeller series and diameter were left unchanged. The parameters of the upgraded propulsion system are shown in table 1.

4. RESULTS Fuel consumption-vessel speed diagrams are presented below. Pusher propulsion systems with theoretically matched pitches, gear ratios and propeller types were analyzed. All the calculations were performed for waterway depth h=4m and draught T=1.6m (corresponding to the fully loaded push train).

4.1 Different pitches Figures 3 and 4 show fuel consumption versus vessel speed for different pitches (the other propulsion system parameters were left unchanged). In the real propulsion system the pitch (P/D) amounts to 0.829. For approximate comparisons P/D=0.5-1.3 with a step of 0.2 was assumed.

Figure 3. Comparison of fuel consumption versus vessel speed for different pitches.

Figure 4. Comparison of fuel consumption versus vessel speed for different pitches.

It appears from the above diagram that the differences in fuel consumption between the propulsion systems with different pitches are slight and difficult to evaluate due to the intersections of the curves. Only narrow ranges of speed in which one pitch generates the lowest fuel consumption can be determined. For this reason no optimum (from the fuel consumption point of view) pitch can be definitely determined for a wide range of vessel speed. Nevertheless, one can notice that the curve for P/D=0.829 is the lowest of the curves shown for the speed of 9-9.5 km/h. As pitch is increased, the propulsion system operating conditions change and higher vessel speeds become optimum. Besides fuel consumption, also the conditions in which the push tug is to operate determine the choice of pitch. Figure 5 shows a comparison of fuel consumption versus vessel speed for different pitch values close to the real pitch, i.e. P/D=0.82 and P/D=0.84.

Figure 5. Comparison of fuel consumption versus vessel speed for different pitches.

It appears from fig. 5 that the curve illustrating fuel consumption for the real pitch runs between the ones for P/D=0.82 and P/D=0.84. This means that for propeller pitch P/D=0.82 and P/D=0.84 fuel consumption is between the values for the other presented pitches, i.e. averaged fuel consumption is ensured. If follows from the above examples that P/D=0.829 is optimum for a vessel speed of 8-11 km/h.

4.2 Different types of propellers Another propulsion system parameter having a bearing on fuel consumption is screw propeller type. Figure 6 shows fuel consumption-vessel speed curves for the Ka 4-55 propeller and other propellers of type Ka.

Figure 6. Comparison of fuel consumption versus vessel speed for different types of propellers.

It appears from fig. 6 that in the prescribed conditions propeller Ka 4-55 generates similar fuel consumption as that generated by propeller Ka 5-75. Moreover, the fuel consumption values for the propeller used in Bizon pusher tugs are higher than the other fuel consumption values in the whole analyzed range of vessel speed. The comparison of propeller types included B-Wageningen series propellers with the same diameter as in the Bizon III pusher tug (D=1.26m) and with D=1.35m. The results are shown in fig. 7.

Figure 7. Comparison of fuel consumption versus vessel speed for different types of propellers.

According to the curves shown in fig. 7 the propeller used in the investigated vessel generates lower fuel consumption than the B-Wageningen propellers.

4.3 Different gear ratios Figure 8 shows the fuel consumption-vessel speed relation for propulsion systems with different gear ratios. Gear ratios: i=3, i=4.5 and i=5 were assumed.

Figure 8. Comparison of fuel consumption versus vessel speed for different gear ratios.

When the gear ratio is changed, the optimum system operating conditions also change. Considering only fuel consumption, the actual gear ratio does not guarantee the lowest fuel consumption. It appears from fig. 8 that higher gear ratios result in low fuel consumption in the whole analyzed range of vessel speed.

5. CONCLUSIONS The above analysis shows that most of the parameters were properly matched with regard to fuel consumption minimization. The correctness of a propeller pitch choice can be assessed only when the navigation conditions in which the vessel is to operate are taken into account. The selected pitch is proper since it ensures the lowest (for the compared pitches) fuel consumption in the prescribed hydrotechnical conditions at a vessel speed of about 9 km/h, which corresponds to the conditions in which Bizon III pusher tugs are actually operated. The analysis of the different screw propellers shows that nozzled propellers are the most advantageous. It appears from the comparison of the series Ka propellers that the three-blade propeller would generate the lowest fuel consumption and the actually used propeller generates the highest fuel consumption in the whole analyzed range of speed. This means that the chosen type of propeller is not optimum in this regard (probably economic considerations won out). The upgraded propulsion systems gear ratio does not ensure the lowest fuel consumption. Higher gear ratios than the actual one will result in lower fuel consumption. But a change of the gear ratio would entail a change in the operating conditions of the whole propulsion system. The analysis shows that the from the fuel consumption point of view the propulsion system parameters were correctly matched.

ACKNOWLEDGEMENTS Paper developed within realisation of research-development program No 10-0003-04 fund by National Centre of Research and Development

REFERENCES Kulczyk J., Sradomski W., Tabaczek T., (2006), Optimization of inland vessel multipropeller propulsion system operating parameters. In: Computer Systems Aided Science. Industry and Transport.' Zakopane TRANSCOMP 2006; River Shipyard in Tczew, (1975), Oder River pusher tug BIZON III-P Specifications (in Polish), Wrocaw, Inland Waterway Transport Research & Design Centre; Kulczyk J., (1986), Fuel consumption in inland navigation conditions (in Polish), Przegld Komunikacyjny No. 6/86, pp.122-125; Nawrocki S., (1965) Model studies of the resistance of Bizon push trains, Gdask, Gdask Polytechnic, Naval Institute;

CONTACT DETAILS Jan Kulczyk, Hydraulic Machines and Systems and Inland Vessels Modelling Division of Wroclaw University of Technology, Wrocaw, Poland Jan.Kulczyk@pwr.wroc.pl

Numerical simulation of the flow around ship and rotating propeller

Tomasz Bugalski, Pawe Hoffmann Centrum Techniki Okrtowej S.A., Gdask, Poland

One of the current research projects of CTO S.A. is focused on investigation of flow around ship hull and rotating propeller. The goal of the project is to test and demonstrate the capabilities of RANSE in connection with complex ship flows. Focus is on a complete CFD model for hull, propeller and rudder, which can account for the mutual interaction between the components when the flow field is calculated. In the present work the real propeller geometry is modelled. The computations required to use sliding grid method a mesh divided into two regions connected with interface surface. First region was a cylinder built around the propeller. Second one was containing the rest of computational domain. The mesh around the propeller rotates relative to a fixed mesh connected with the hull. In order to check the performance of the CFD model, the calculated results are compared with experimental data. The paper presents the results of preliminary investigation of the flow computations around Nawigator XXI training ship and ITTC KRISO Container Ship KCS. All meshing and flow simulations are conducted with StarCCM+ from CD-adapco.

Keywords: ship hull propeller interaction, wake flow, RANSE

1. INTRODUCTION Hydrodynamic aspects play an important role in the quality of a ship. Dominant criteria in the hull form design are often the resistance and powering performance; in addition, the occurrence of noise and vibrations, important for the comfort level of crew and passengers, often has a hydrodynamic cause, stemming from the operation of the propeller in the disturbed flow field behind the ship hull. Therefore, the detailed knowledge of the flow around the aft-part of a ship is critical: In this area, flow separation may occur, with important consequences for the resistance and power; and the flow field felt by the propeller should be determined. Unfavourable characteristics of this flow field (e.g. large spatial variations of the velocity) may result in performance loss, propeller cavitation leading to noise, vibrations and erosion, etc. The viscous flow can be computed using Reynolds Averaged Navier Stokes (RANSE) codes. This is becoming a common component in the ship design process, and good agreement with model-scale wake field measurements can be achieved today. The real purpose is to predict the hull/propeller interaction, wake field and finally full-scale flow about ship hull. RANSE computations offer that possibility, and such viscous-flow computations start to be used in the practical ship design; but how accurate these predictions are, and to what extent they depend on the turbulence modelling used, is not really known. A validation of the full-scale ship viscous flow predictions has generally been insufficient so far, mainly due to the virtual absence or difficult accessibility of the suitable full-scale experimental wake field data. The results of preliminary numerical simulation of the flow around ship and rotating propeller are presented below. The computations were carried out at model scale, to enable direct comparison of the CFD results with the experimental results.

2. DESCRIPTION OF THE COMPUTATIONAL OBJECTS In this part of paper, main data of the analysed objects as well as the ship and propeller geometry and tested configurations are presented. The non-standard model tests, for training/research vessel Nawigator XXI have been performed at the Ship Hydromechnics Division of Ship Design and Research Centre S.A. (CTO S.A.) in the frame of the European-Union EFFORT (European Full-Scale Flow Research and Technology) project. The training/research vessel Nawigator XXI is operated

by the Maritime University of Szczecin. The study was conducted on both the full ship and the model scale. The KCS was conceived to provide data for both explication of flow physics and CFD validation for a modern container ship with a bulbous bow (i.e., ca. 1997). The Korea Research Institute for Ships and Ocean Engineering (now MOERI) performed towing tank experiments to obtain resistance, mean flow data and free surface waves (Van et al, 1998a,b, Kim et al, 2001). Self propulsion tests were carried out at the Ship Research Institute (now NMRI) in Tokyo and are reported in the Proceedings of the CFD Workshop Tokyo in 2005 (Hino, 2005). Data for pitch, heave, and added resistance are also already available from Force/DMI measurements reported in Simonsen et al. (2008). No full scale ship exists. Main characteristics of the hull shapes, scale factors and visualizations of the hull shapes are presented below in Table 1.

Table 1 Main data of the ships

Ship Length b.p. [m] Length of waterline [m] Breadth [m] Draught [m] Displacement [m3] Surface wetted area [m2] Block coefficient Midship section coefficient Service speed [kn] Propeller Direction of rotation Diameter [m] Number of blades Expanded area ratio Pitch ratio Blade rake Skew back [deg] Profile

Training ship NawigatorXXI 54,13 55,16 10,50 3,15/3,20 1126 672 0,623 0,915 9,98 CP469 left 2,26 4 0,6728 0,9424 0 28 NACA16/0.8

Container ship KCS 230,00 232,50 32.20 10,80 52030 9424 0,651 0,985 24,0 KP505 right 7,90 5 0,7963 1,1740 0 0 NACA66/0.8

As mentioned above, the computations were carried out at model scale. The scale factors are the same as for the models used for towing tank experiments, and are listed in Table 2. The scaled values of speed are also given.

Table 2 Scale factors and speeds in model scale

Ship Scale factor Model speed [m/s]

Training ship 10,00 1.609

Container ship 31.60 2.196

Figure 1 Geometry of the hulls and propellers training/research ship (left) and container ship (right)

3. METHODOLOGY OF THE COMPUTATIONS The computations are performed with the Reynolds Averaged Navier-Stokes (RANSE) solver StarCCM+ from CD-adapco. The code solves the RANSE and continuity equations on integral form on a polyhedral mesh by means of the finite volume technique. Both steady state and transient calculations are considered. The Reynolds stress problem is solved by means of k- turbulence model. The rotating propeller is treated in different ways. For open-water calculations the propeller inflow is uniform, so the moving reference approach is applied. For the propeller rotating behind a ship, a rigid body approach is applied. The free surface is modelled with the two phase volume of fluid technique (VOF). The following jobs were developed and performed in this research project: development of discretization method for hull propeller system, development of propeller hydrodynamic characteristics computations using moving reference frame,

comparative calculations of open water propeller (equivalent for open water propeller test) and flow around ship hull without propeller (equivalent to resistance test) were made,

calculation of flow around ship with working propeller was made (equivalent for propulsion test) was made,

Figure 2 Computational domain in open water propeller test.

The CFD computations of the flow around the ship hulls with rotating propeller were carried out with the use of RANSE flow model, implemented in STAR CCM+ solver. The basic assumptions for the simulation were as follows: the free surface was neglected in order to reduce the convergence time and to improve stability of the computations, the dynamic trim and sinkage of the hulls were also neglected, propeller operation was modelled. The unstructured computational mesh of hexahedral and polyhedral cells was used for the computations. The total number of elements used in the calculation of the grid was approximately 1.8 million. The visualizations of mesh details are presented in figures. Computational domain consists of two parts the cylindrical volume in which there is a propeller (Fig. 2) a separate sub domain for propeller calculations (semi-transparent outer wall.)) and the rest of the volume - separated by intermediate surface (interface) that is providing a connection between boundaries during the simulation. A larger volume is divided into a grid of hexahedral cells, which size and number is adjusted to the complexity of the flow in the area. The area near the propeller is of particular interest, therefore the mesh was refined there. A smaller cylindrical region connected to propeller is generated as a grid of polyhedral cells which is more suitable for mapping the shape of a rotating propeller.

The solver settings for the simulation were as follows: turbulence model: k -epsilon time step: 0.001 s; number of iterations per time step: 3.

Figure 3 Mesh in the aft part of the ship hull

4. RESULTS OF CALCULATIONS The open water calculation was carried out at the same running conditions as used in the experimental set up at CTO S.A. and KRISO. The solution domain was chosen to extend 10 propeller diameters in front of the propeller, 3 diameters in the radial direction and 5 diameters behind the propeller. The flow solver was run in steady mode and the rotation of the propeller was accounted for by using the moving reference frame approach. This approach works fine in open water, where the propeller experiences a completely uniform inflow field. The calculation was carried out for advance coefficients in the range from J=0.0 to J=1.0. Similar to the experiment the computed thrust and torque on the propeller were converted into the dimensionless thrust coefficient, torque coefficient and the efficiency was calculated. The study of the flow field shows that the propeller accelerates the flow over the blades and introduces swirl in the flow downstream of the propeller, as expected. The pressure field on the blades shows low pressure on the suction side and high pressure on the pressure side. Following Figure 4 illustrates the pressure distribution on the suction side of both tested propellers.

For turbulence modelling the k-epsilon turbulence model was used. For movement simulation the moving reference frame physical model was used. Calculations lasted about 2 days using a single processor for one point in advance coefficient characteristics J (V, n).

Figure 4 Pressure distribution on suction side of propellers (open water test conditions)

Figure 5 Comparison of open water hydrodynamic characteristics of P505 propeller computed and obtained from experiment.

The results of simulation of open water tests for CP469 propeller model (Nawigator XXI) were the basis for positive evaluation of moving reference frame method. In the comparison between the calculated and measured data for P505 KCS propeller, it is seen that fairly good agreement is achieved. Particularly in the region around J=0.6-0.8, which is the point where the ship usually operates. The goal of the present project is to simulate the flow about complete ship with hull, rudder and propeller including the free surface around the ship. The simulation with the propeller behind the ship must to be run in transient mode, i.e. time accurate. In order to verify the reliability of the CFD simulations, the flow about model of training vessel Nawigator XXI trusted by CP469 propeller was computed, and the results were compared with the existing experimental results. The computation results are presented in following figures.

Figure 6 Pressure distribution on the working CP469 propeller right side view

The study of the field quantities, velocities and pressure in the stern region shows a time varying, periodic flow field, which is related to the blade frequency of the rotating propeller.

Figures 6, 7 and 8 show an example of the pressure field in the stern region and the axial velocity contours in a cross section at the rudder position. With respect to the velocity field, it shows that the propeller accelerates the flow and introduces swirl in the flow downstream of the propeller.

Figure 7 Axial velocity field in symmetry plane for Nawigator XXI, CP469 propeller and R588 rudder (model scale)

Figure 8 Axial velocity field (total) in front of working propeller (Nawigator XXI, CP469 propeller)

Figure 9 Pressure distribution on the rudder at zero degree position and propeller at twelve oclock blade position (port side and starboard view)

Figure 10 Calculated pressure on the suction and pressure side of CP469 propeller at twelve oclock blade position (starboard side) or downward (port side)

As can be seen the wake of the ship strongly influences the propeller, but the propeller also influences the ship flow. The influence is most clearly seen in the pressure field. Upstream of the propeller the hull experiences suction, which reduces the pressure and increases the resistance. The effect is usually expressed as the thrust deduction. Another region of the hull that feels the presence of the propeller is the region above the propeller. In this region the passing blades will introduce pressure pulses on the hull, which in critical cases can lead to noise and the ship structure vibration.

5. CONCLUSIONS Planned results of research task including: development of discretization method for rotating propeller using STAR-CCM+ solver, development of computation method for hull propeller system and obtaining reliable results of the calculation of flow around a rotating propeller and viscous flow around a model ship with working propeller have been achieved. Calculations of flow around the ship hull with rotating propeller simulating propulsion tests have an unsteady character and therefore need more time to obtain convergent result. Calculations using sliding grids as in hull propeller system are characterized by lower stability compared to simple flow computations hence using of lower relaxation factors and higher number of iterations per time step is necessary. As a result the calculation time increases significantly and the convergent result is more difficult to obtain. At the same time the opportunity arises to analyze the phenomena occurring locally in the time domain. For the practical usefulness of the method, accuracy in determined suction coefficient should be close to accurate determination of the resistance.. The method of calculating using the grid is already articulated in the continued application of research projects carried out under the CTO SA, followed by development methods and a thorough verification of the quality of its operation. After fine tuning and a broader review, including the confrontation within the international benchmark Gothenburg2010, the method is applicable to the tasks of numerical analysis and commercial buildings with complex geometry.

ACKNOWLEDGEMENTS The research presented in this paper has been financially supported by The Polish National Centre for Research and Development (NCBiR), grant No. N R10 0039 06/2009 and Ministry of Sciences and Higher Education (MNiSW), internal statutory grant of CTO S.A. No. A3:800-14/2009. The authors would like to express their gratitude for this support.

REFERENCES Bugalski T., (1997), Modern Methods for Investigation of Hull -Propeller Interaction Phenomena using CFD. NAV&HSMV International Conference , Sorrento, pp. 2.3-2.8. Bugalski T., Kraskowski M.: (2006), Validation of the RANSE wake computations for the training ship Nawigator XXI and the dredger Uilenspiegel, Proceedings of 9th Numerical Towing Tank Symposium (NuTTS), Le Croisic, France, pp.4.1-4.6. Bugalski T., Szantyr J., (1998), Application of CFD for Analysis of the Ship and Propeller Flow. TASK Quarterly 2/98 Scientific Bulletin of the Academic Computer Centre in Gdansk. Menter F., Maksoud M.-A., (2002), Practical numerical simulation of viscous flow around ships. The Naval Architect, pp. 36-40. Simonsen C.D., Carstens R., (2008), RANS Simulation of the Flow around a Ship Appended with Rudder, Ice Fins and Rotating Propeller, Proceedings of 11th Numerical Towing Tank Symposium (NuTTS), Brest, France. User Guide STAR-CCM+ Version 4.02.011, (2008), CD adapco. Wake-Fields Committee Final Report and Recommendations to the 25th ITTC, (2008), (Bugalski T., co-author). Van, S.H., Kim, W.J., Yim, D.H., Kim, G.T., Lee, C.J., and Eom, J.Y., (1998a), Flow Measurement Around a 300K VLCC Model, Proceedings of the Annual Spring Meeting, SNAK, Ulsan, pp. 185-188. Van, S.H., Kim, W.J., Yim, G.T., Kim, D.H., and Lee, C.J., (1998b), Experimental Investigation of the Flow Characteristics Around Practical Hull Forms, Proceedings 3rd Osaka Colloquium on Advanced CFD Applications to Ship Flow and Hull Form Design, Osaka, Japan.

CONTACT DETAILS Tomasz Bugalski, Centrum Techniki Okrtowej S.A., ul. Szczeciska 65, Gdask, Poland tomasz.bugalski@cto.gda.pl Pawe Hoffmann, Centrum Techniki Okrtowej S.A., ul. Szczeciska 65, Gdask, Poland pawel.hoffmann@cto.gda.pl

Design of Wind Tunnel for Marine Applications

Leszek Wilczyski Ship Design and Research Centre S.A., Gdask, Poland

The paper contains general considerations regarding design of low speed wind tunnel for marine objects model testing. The discussion of principles and methods applied during the design process in connection to the assumed operational envelope of the tunnel is presented. The most significant features of the discussed example of the wind tunnel are: the area of test section entrance transverse cross section 12m2, maximum air flow velocity 50 m/s. The paper concerns primarily design and performance of the closed circuit wind tunnel with closed jet test section, however substantial part of considerations can be applied directly during designing of slotted, open jet or even open circuit tunnels. The paper presents the overview of the tunnel performance prediction and analyzing methods suitable exclusively for the low speed tunnels. It means that the air compressibility is neglected, what reflects to the assumption that the maximum air velocity does not exceed 0.3 Ma. Keywords: low speed wind tunnel, model testing, ship, off-shore

1. INTRODUCTION Despite the significant progress in computational fluid dynamics methods a low speed wind tunnel (LSWT) is very often considered as an ultimate source of data regarding aerodynamics of various objects. LSWT is a research facility which allows for controlled simulation of the air flow around a tested object (or set of objects) and determination of aerodynamic forces acting on it. The flow velocity at the entrance to the test section (TS) is the principal physical

quantity, that is controlled during the tests in LSWT and determines the range of applicability of the tunnel. A scaled model of real object or its part is the most often tested physical sample. As long as the size of the test section is sufficient to accommodate full scale object (i.e. the TS blockage coefficient is acceptable as well as the tested object wake does not affect the operation of the LSWT) then such objects as passenger cars, trucks, motorcycles, antennas, solar panels, parts of ship deck equipment, sportsmen or even small (e.g. remotely operated) aircrafts are tested. The primary goal of tests conducted in the wind tunnel is to obtain the knowledge about aerodynamics of particular object on the basis of experiments performed either using a scaled model or a prototype. Sometimes when the object has got clear symmetry plane (as in case of an airplane or a ship) its half model is tested. The rough test result, requires error analysis and appropriate scale effect interpretation. According to the above, the most important problem connected with aerodynamic tests in a tunnel is to guarantee similitude between the flow simulated during the experiment and real one, occurring around full scale object. Another significant challenge of the aerodynamic model testing is the air flow quality in the test section. It is important to ensure the high level of the homogeneity of the inflow to the tests section, lack of the flow separation in any section of the tunnel (especially diffusers) low level of turbulent velocity fluctuations and the smallest possible longitudinal component of the pressure gradient in the test section. The achievement the flow similitude requires the identity of tested objects geometry (also in scale) and equality of dimensionless similitude numbers determined for a simulated flow as well as for real flow. In case of steady objects tested in the LSWT, for example prototypes or models of aircrafts, cars, bridges, buildings, above water parts of ships or drilling rigs (what provides the great majority of all experiments) keeping the identity of Reynolds number Re is of the greatest significance. The most often the compressibility of the air in LSWT is neglected (not exceeding 1% when Mach number Ma 0.4) and it is presumed that the dominating interactions for flowed body are viscosity forces. 2. BASIC CHARACTERISTICS OF THE DESIGNED WIND TUNNEL The essential features of the designed tunnel are as follows: The tunnel has got a form of horizontal closed, return, rectangular, solid wall circuit. The test section is a closed jet type with solid walls. Along the circuit particular sections can be distinguished. The division of the tunnel into sections and their nomenclature is presented in Fig 1 and Tab.1. Except the drive unit the transverse cross-section of the air flow circuit is kept rectangular. The cross-section aspect ratio is kept close to the value W/H= 1.5 along the entire construction. This particular

feature of the tunnel internal shape has been set in order to reduce the risk of the accidental vortex generation in the vicinity of transverse cross-section corners. There are 4 elbows along the circuit, each turning the air flow by 90, each equipped with a set of vertically oriented, aerodynamically shaped guide vanes. The horizontal cross-section of the individual vane if based on two arcs, which are tangential to the circle at the leading edge and intersect at the trailing edge. Two high speed elbows are mirror images. The same concerns the pair of low speed elbows. The guide vanes in each elbow are distributed unevenly along its diagonal. The distance between vanes increases with the distance from the internal bend. The shape of the vanes in high speed elbows is kept identical regardless its location. The same concerns the low speed elbows, however high speed and low speed vanes are different. The drive unit of the tunnel is composed of 6 parallel, identical, single stage axial fans. Each fan characterizes with the same air flow rate and total pressure rise generated at certain revolutions. Revolutions of each individual fan can be altered smoothly. The drive unit is composed of two horizontal layers, each containing 3 fans as presented in Fig. 1. This solution reflects to the principal design concept consisting in keeping the tunnel cross-section aspect ratio constant. Each fan duct is coupled with its individual contraction and diffuser. Each single fan is driven individually by the electric motor which itself is shielded by the aerodynamically shaped housing. The internal volume of each electric motor shield is cooled by the air which is sucked from outside and released outside of the tunnel. The static pressure in the test section is balanced with the ambient static pressure (atmospheric). There are two sets of breathers located in the area of connection between the test section and the first diffuser on both tunnel side walls. This solution is very convenient from the tunnel user point of view. Small holes in the test section walls do not introduce significant disturbances to the measurements. To reduce the risk of the pressure gradient (due to the expected growth of a boundary layer and reduction of the hydraulic diameter) vertical walls of the test section diverge with small angle with respect to the section horizontal axis. Top and bottom walls of the test section remain parallel. This particular feature of the tunnel reduces the risk of occurrence of the so called horizontal buoyancy. There is a safety screen preventing drive unit from accidental damages. The screen is installed just in front of the leading edges of guide vanes in the second high speed elbow. The contraction coefficient of the contraction section of the tunnel equals 7. This is a commonly accepted value, still obtainable in the tunnels without wide angle diffusers.

One of the design indications of the tunnel was to reduce as much as possible the risk of any flow separation. Therefore the wide angle diffuser has been excluded from the application and two main diffusers are relatively long with a small divergence angle. Between a second low speed elbow and a contraction there is a section of a constant cross-section area, containing a honeycomb and two turbulence reducing screens. The honeycomb cells are hexagonal. The screens are the wire nets of rectangular mesh. Both screens are tensioned and the tensioning mechanism remains outside of the air flow anytime. There is a cooler in a section of constant cross-section area between a second diffuser and a first low speed elbow. The cooler is composed of sets of vertical pipes equipped with a flat plate tail and a set of horizontal plates. One of the additional features of the cooler is to equalize the wind velocity profile at the second diffuser outlet. The corners of the test section, contraction and first diffuser are filled.

The primary operational characteristics of a LSWT are as follows: Maximum speed of the undisturbed flow at the entrance to the test section 50 m/s Maximum stagnation pressure of the air the test section is balanced with ambient pressure Air temperature assumed to be kept constant (app 40 C) during the tunnel operation, thanks to the use of an internal cooling installation. Pressure gradient along the axis of the TS to be less than 1% of dynamic pressure, Turbulent fluctuation of the velocity at the entrance to the TS to be less than 0.5 % of maximum velocity.

Figure 1. General layout and division of the LSWT into sections

Table 1 LSWT - sections numbers, names and symbols.

1 2 3 4 5
6 7

Section Test section First Diffuser Connector 1 High Speed Elbow 1 High Speed Leg
Safety Screen High Speed Elbow 2

Symbol TS D1 C1 COR1 CYL1

SS COR2

11 12 13 14 15
16 17

Section Cooler Section Low Speed Elbow 1 Low Speed Leg Low Speed Elbow 2 Honeycomb
Turbulence Reducing Screen 1 Turbulence Reducing Screen 2

Symbol CS COR3 CYL2 COR4 H

S1 S2

8 9 10

Connector 2 Drive Unit Second Diffuser

C2 DU D2

18 19 20

Settling Chamber Contraction Connector 3

SC CON C3

3. METHODOLOGY OF THE TOTAL PRESSURE LOSSES CALCULATIONS The choice of the appropriate drive unit of the tunnel requires the calculation of total pressure losses (flow losses) conducted for the entire circuit. The main features of the applied method are as follows: The entire circuit of the tunnel is divided into distinguishable sections, according to their shapes and functions. The total pressure loss and local flow losses coefficients for each section are determined. Then the relative flow losses coefficients related to the test section dynamic pressure for each section are determined. Hence the tunnel is a low speed one, thus the air compressibility is neglected and consequently the flowing medium is considered as the viscous, incompressible fluid. It is assumed that the temperature of the circulating air remains constant as well. The elements of internal tunnel structure are taken into account as contributing to the total pressure losses. The following parts of the tunnel and simplifying assumptions regarding their influence on the air flow are considered: o o o safety screen is modelled individually as bars, guide vanes are taken into account the contribution of the honeycomb cells are taken into consideration,

o the turbulence reducing wire screens are modelled with no major simplification regarding their geometry,

o drive unit is modelled as the set of six ducts, each equipped with contraction and diffuser, the influence of the motor gondola on the total pressure losses is neglected The power required to maintain constant and steady air flow through the test section with the certain discharge equals to the total flow losses occurring in the entire tunnel circuit. These loses are due to the kinetic energy dissipation caused by air viscosity (vorticity and turbulence). The loss in the kinetic energy of the air stream, which appears as the decrease in total pressure must be compensated by a pressure rise, provided by the set of 6 single stage axial fans. Thus denoting the average fan efficiency by , the energy balance equation can be written as follows:

P = energy dissipation time rate in i th tunnel sec tion

What leads to the following

P = k i 0
i

where pressure losses coefficients in i-th section can be written as (q denotes dynamic pressure): p q ki 0 = i i . qi q0 4. CONCLUSIONS The correctness of the shape of the designed LSWT has been analysed numerically with the use of CFD methods and experimentally, on the way of model testing at the scale of 1:3. Both methods provided promising results regarding the maximum achievable flow speed, velocity field and the lack of the flow separations. The combined use of numerical and experimental approach seems to be the most efficient way of the design of such facilities like large scale LSWT. ACKNOWLEDGEMENTS The paper has been prepared within the frames of the research project no. 6 ZR6 2009 C/07216 Tunnels for testing of the wind interaction on land and marine objects supported partly by Polish Ministry of Science and Higher Education. CONTACT DETAILS Leszek Wilczyski, CTO S.A., Rzeczypospolitej 8, Gdask, Poland Leszek.Wilczynski@cto.gda.pl

Mathematical Model and Numerical Calculations of the Pelagic Trawl System

Czesaw Dymarski, Jacek Nakielski Technical Univeristy of Gdask, Gdask, Poland

This paper presents a mathematical model of the mass centre motion trajectory of the trawl, its main geometrical parameters, way of solving the model as well as its example results. It could be useful to define optimum parameters of safe, effective and environmentally friendly pelagic fishing.

Key words: sea fishing, trawl fishing, physical and mathematical model of trawling

1. INTRODUCTION Exploitation of sea life resources undergoes continuous changes which are aimed at fish catch increase on one hand and protection of the resources and ensure of biological balance to marine ecosystems on the other hand. Various forms of reducing unfavourable consequences of the exploitation are applied a.o. by limiting catch of particular fish kinds as well as by prohibiting application of harmful fishing techniques and preferring environmentally friendly ones. One of the relatively environmentally friendly fishing method is pelagic fishing with controlled position of trawl in sea depth. The paper presents a mathematical model and a way of its solving for calculation of the mass centre motion trajectory as well as geometric parameters of the trawl during pelagic fishing. In the mathematical model there are taken into account speed, pitch and heaving of the ship, as well as velocity of running out (or hauling) the line by trawl winch and also a shape and dimensions of the trawl system. In the paper presented also same examples of calculation results. Elaborated mathematical model and based on it computer program can be very useful tool for determination of optimal parameters of trawling what is necessary to carry out highly effective pelagic fishing.

2. ASSUMPTIONS FOR PHYSICAL MODEL OF THE TRAWL A fishing cutter together with laid down trawl is presented in a simplified way in Figure 1.

Figure 1. A simplified image of fishing trawl gear.

The trawl gear is consisted of trawl warps which are wound around drums of two trawl hoisting winches. The lines go into water through pulley blocks fastened to the aft frame mast. Deep in the water they are fixed to the trawl boards by which an appropriate divergence (width) of the trawl net is obtained. From the trawl boards four trawl legs go. Two upper ones are mutually connected through the floatline, two lower ones - through the bottom line. The side legs are connected in pairs through two side lines. To the floatline the floats are connected , due to which an appropriate floatability of the gear is obtained, and the bottom line is loaded, additionally onto the lower trawl legs the main weights are hung. Due to the buoyancy force and applied weights the trawl mouth is shaped in vertical plane. The next element of the trawl gear is its net part consisted of a few conical segments ended with a cylindrical tail into which caught fishes fall.

Figure 2. Simplified model of trawl gear whose particular numbered parts are substituted by their masses concentrated in their gravity centres, namely : 1 trawl net part; 2 floats; 3 weights: 4 trawl boards; 5 point of line run-out from the ship; G, R, F gravity , buoyancy, drag and line stretching forces, respectively.

One of the more important problems which occur during pelagic fishing is to appropriately select operational parameters of ship and trawl winches so as to ensure keeping the trawl in operation in a given water depth. So far not many fishing ships have been equipped with instruments to control trawl position hence choice of the above mentioned parameters is mainly based on crew experience. However not always that will do. Many fishermen resign from carrying out such kind of fishing for fear of catching the trawl on various obstacles placed on the sea bottom and loss of the trawl net together with caught fishes. The situation

justifies purposefulness of elaboration of a mathematical model and calculation software based on it , which could make it possible to determine trajectory of trawl motion in water during fishing, depending on main operational parameters. Exact calculation of motion trajectories of particular elements of trawl gear immersed in water is a very difficult task not only with a view of possibility of solving the complex and mutually confounded equations but also due to difficulties in defining and describing randomly changeable sea conditions. For this reason such calculations are to be performed under certain, often very far going, simplifying assumptions depending on a given aim and expected accuracy of results to be obtained, resulting from the aim. Below is presented a relatively simple model which makes it possible to determine motion trajectory of trawl gear gravity centre during fishing in waves and at changeable operational parameters of fishing ship or trawl hoisting winch. To the model the following simplifying assumptions were introduced to describe the above presented trawl gear: - the trawl gear is symmetrical with respect to ship plane of symmetry , that means that both trawl warps and the part of the gear, connected with them, geometric, kinematic and dynamic parameters ; - the following parts of the gear , namely : trawl net, floats, weights, trawl boards and warps are assumed to be finite material points placed vertically in estimated gravity centers of the parts ; - the above specified material points have masses and drag areas equivalent to those represented by relevant parts of the gear, and they are connected each to other by means of straight line segments; - motions of the ship , trawl warp and trawl net in waves occur in vertical plane of ship symmetry; - trawl warps are assumed to be uniform, perfectly flexible and inextensible strings with taking into account their deflection due gravity and buoyancy forces and hydrodynamic interaction due to their motion in sea water ; - the origin of Cartesian coordinate system is located in the point of contact of the trawl warp with ship stern in the place of going the warp down to water. Schematic diagram of the trawl gears physical model which satisfies the above specified simplifying assumptions, is shown in Figure 2 and 3. are of identical

3. MATHEMATICAL DESCRIPTION OF THE ASSUMED MODEL To calculate motion trajectory of the trawl gear the equations of component forces applied to its gravity centre can be described as follows:
Rw = Ri - total horizontal drag force
i =1 i=4

(1)

Gw = Gi - total gravity force in sea water

i =1

i=4

(2)

where: Ri =
2 ix

C ix Aix

(3)

sea water density, Cix, Ciz drag coefficients of i-th element of the trawl gear in x - and z axis direction, respectively, Aix, Aiz projected areas of i-th element in x - and z axis direction, respectively , Gi, weight (buoyancy) of i-th element in sea water; ix, iz velocity of i-th element in x - and z axis direction, respectively:

ix = 5 x + l cos l1

(4)

iz = 5 z + l sin l1

(5)

l velocity of running out + ( or hauling -) the line by the trawl winch; l1 = arctan(Gw / Rw) the angle between the line and x- axis at its connection with the
trawl net 5x speed of the ship (and the point 5) in x axis direction, assumed constant; 5z speed of the ship stern (and the point 5) in z axis direction, under the assumption that motion of the point 5 due to heave and pitch motions of ship in waves is close to sinusoidal, hence it can described by the equations : z5 (t ) = a5 z sin( t ) (7) x5 5 x t (8) (6)

a5z displacement amplitude of the point 5 in z-axis direction,

= 2 / T
where: T ship pitch period t time.

(9)

Calculation of shape of the trawl warp in vertical plane was performed by using the numerical method in accordance with the model presented in Figure 3.

Figure 3. Simplified model of shape of the trawl warp ( line) , taken to numerical calculations

The force in the end of the trawl warp at the point of its connection with the trawl body is equal to :

Fw = Fl .1 = Rw 2 + Gw 2
and, in an arbitrary point of the line it can be calculated from the relation:

(10)

Fl i = Fl i 1 + Fl i
where:

(11)

Fl i 1 = Fw + Flk
k =2 2 Fi = Rs Xi + (Gs Rs Zi ) 2

k =i

(12)

(13)

Gs weight of the calculated segment of the line in sea water

Rs Xi = Rsi sin 2 ( si + i ) (14) RsZi = Rsi cos 2 ( si + i ) (14)

si average slope angle of the calculated segment of the line , calculated from the
following interpolation :

si = 1,5 i 1 0,5 i 2
i-1, i-2 slope angles of the line in the points i1 and i2 , respectively,

(14)

l1 = arctan( Rsi / Rsni )

(15)

2 Rsi = Rsni + Rsi

resultant drag force of the calculated segment of the line

(16)

2 Rs ni = 0,5 Cn x d S sin 2 si drag force of the calculated segment of the line in

the direction perpendicular to it

(17)

2 Rsi = 0,5 C x d S cos 2 si drag force of the calculated segment of the line in

the direction parallel to it where : Cn drag coefficient of the line in the direction perpendicular to it , C drag coefficient of the line in the direction parallel to it, d diameter of the line S length of the elementary ( calculated) segment of the line.

(18)

Coordinates of an arbitrary point i of the trawl warp relative to the point of its connection with the trawl body were determined from the following relations : xiI = xiI1 + S cos Si z iI = z iI1 + S sin Si and the slope angle of the line from the interpolation formula: (19) (20)

li = li 1 + 2 ( Si 1 li 1 ) = 2 Si 1 li 1

(21)

Any motion of the point 5 in z axis direction will produce definite changes in loads in the line and displacements of the gravity centre of the trawl gear. The equation of forces acting on the trawl gear in z axis direction takes the following form:

d 2z dz i = 4 = mg Fz (Ciz Aiz ) 2 dt 2 dt i =1
2 i =4 i =1

(22)

where: m = mi total mass of the trawl gear elements in sea water (23)

I z = z n vertical coordinate of the trawl gear gravity centre, relative to the point 5 (24)

l = l0 + l t current length of the run-out trawl warp; Fz = R tg l1 vertical force component in the line , acting on the trawl gear;

(25) (26)

To solve the differential equations Runge-Kutta method was used. At first the following substitution was introduced:
y= dz dt

(27)

On substitution of Eq. (27) to Eq. (22) and appropriate transformations the following was obtained :
dy i=4 dz i =4 2 =g C ix Aix ix tg l1 (t ) (C iz Aiz ) 2 m i =1 2 m dt i =1 dt

(28)

The calculation process was realized by applying a constant time step and assuming the results of the preceding step to be initial value for the successive calculation step. To calculate geometric parameters of the trawl gear in steady operational conditions the above given mathematical description was supplemented by the following additional relations to which the notations complying with Figure 4 were introduced:

Figure 4. Simplified model of the trawl gear, with indicated load vectors and main geometric parameters

= arctan

2 (G cos + R sin ) 2 2 2 2 R1 + G1

(29)

In the first calculation step :

= l

= arcsin

l1 sin l2

(30)

If the above mentioned angles and the line length is known it is possible to determine coordinates of gravity centers of particular parts of the trawl gear, namely :
I x4 = x n I z4 = zn

(31) (32) (33) (34) (35) (36) (37) (38)

x3 = x4 l 2 cos( + ) z 3 = z 4 l 2 sin ( + )

x2 = x4 l 2 cos( ) z 2 = z 4 l 2 sin ( ) x1 = x4 l1 cos l 2 cos z1 = z 4 l1 sin l 2 sin

Then, for the so determined geometry of the gear the moment in the point 4 , which tends to rotate the gear with respect to the point , is calculated. M 4 = G1 l14 cos + G2 l 2 cos( ) + G3 l 4 cos( + ) R1 l14 sin R2 l 2 sin( ) R3 l 4 cos( + ) (39) (40)

l14 = l1 cos +l 2 cos

In the case when a calculated value of the moment is greater than an assumed accuracy, for the next calculation step a little greater or smaller value of the angle is taken, depending on sign of the moment. When a sufficiently accurate result is achieved the coordinates of the points 1 through 4 and then the moment relative to the point 5 are calculated for the new geometry of the trawl gear.
M 5 = G1 (l14 cos + x4 ) + G2 [l2 cos( ) + x4 ] + G3 [l 4 cos( + ) + x4 ] + G4 x4 R1 (l14 sin + z 4 ) R2 [l 2 sin( ) + z 4 ] R3 [l 4 cos( + ) + z 4 ] R4 z 4 (41)

Like before, the angle l1 will be corrected depending on value and sign of the moment until its values in successive iteration steps are sufficiently accurate. The computer software in question was written in Fortran programming language to make it possible to carry out calculations in various working conditions.

4. RESULTS OF NUMERICAL SIMULATIONS

Below are presented example results of the calculations of the trawl gravity centre motion trajectory, performed for the following operational conditions: A1x = 30 m2; A2x = 0,471 m2; A3x = 0,028 m2; A4x = 0,422 m2; A1z = 20 m2; A2z = 0,736 m2; A3z = 0,028 m2; A4z = 0,19 m2; G1 = 245,25 N; G2 = -40,42 N; G3 = 245,25 N; G4 = 147,15 N; x = 1,5 m/s and 2,0 m/s; l = 300 m; a5z = 0,5 m; T = 7 s; Cn = 1,2; C = 0,02; d = 0,01 m and three values of rope velocity:

l = 0; 0,06; and -0,06 m/s.

Diagram 1. Trawl displacement velocity for the parameters: s = 1,5 m/s; l = 300 m; a5z = 0,5 m and T = 7 s.

Diagram 2. Trawl displacement for the parameters: s = 1,5 m/s; l = 300 m; a5z = 0,5 m and T = 7 s.

Diagram 3. Trawl displacement velocity for the parameters: s = 2 m/s; l = 300 m; a5z = 0,5 m and T = 7s.

Diagram 4. Trawl displacement for the parameters: s = 2 m/s; l = 300 m; a5z = 0,5 m and T = 7 s.

The oscillations of the trawl gear gravity centre of for 300 m length of the laid-down line and the trawling speed of 1,5 m/s, amount to 0,226 m and along with the increasing of trawling speed they maintain on similar level. And, the displacement velocity of the gravity centre amounts to z = 0,06 m/s on average. In the case of decreasing the initial length of the laiddown line the mean value of oscillation amplitude increased up to 0,289 m and the displacement velocity was equal to almost 0,09 m/s.

It should be also observed that for 300 m length of the laid-down line, for the first several seconds (one period) after switching the winch on line hauling mode, the flattening of the displacement characteristics occurs at simultaneous deviation of its mean value. The

emerging of the trawl occurs relatively fast at simultaneous insignificant increasing the amplitude of oscillation, and an increase of its draught with accompanying decrease of the oscillation amplitude takes place during laying down the line .The trawl velocity vz , after starting the winch and transient disturbances resulting from that, tends to be close to the initial velocity except that , as expected, oscillation amplitude value increases in the case of hauling the line and decreases during laying down the line. The increasing of trawling speed or decreasing of length of the laid-down line does not cause any distinct intensification of the above mentioned phenomena. On Diagrams 3 and 4 show displacement and displacement velocity for vessel speed 2 m/s. Profile of this graphs are very similar like on Diagrams. 1 and 2. Below in Tab. 1. and the diagrams are presented example results of the numerical calculations performed for two values of the laid-down line length l = 300, 200 and 150 m as well as five values of the ship speed s = 1; 1,25; 1,5; 1,75 and 2 m/s.
Table 1. Results of the simulations

x [m/s] 1 1 1 1,25 1,25 1,25 1,5 1,5 1,5 1,75 1,75 1,75 2 2 2

l [m] 300 200 150 300 200 150 300 200 150 300 200 150 300 200 150

l [] 6,451 6,451 6,451 4,139 4,139 4,139 2,877 2,877 2,877 2,114 2,114 2,114 1,619 1,619 1,619

[] 6,138 6,037 5,958 3,941 3,914 3,827 2,740 2,696 2,661 2,014 2,982 1,956 1,543 1,518 1,498

[] 7,834 7,834 7,834 5,064 5,064 5,064 3,530 3,530 3,530 3,597 3,597 3,597 1,990 1,990 1,990

[] 3,168 3,168 3,168 2,051 2,051 2,051 1,430 1,430 1,430 1,053 1,053 1,053 0,807 0,807 0,807

Diagram 5. Calculated values of the horizontal coordinate x of the points 14 for l = 300 m and s =12 m/s.

Diagram 6. Calculated values of the vertical coordinate z of the points 14 for l = 300 m and s =12 m/s.

Diagram 7. Simplified shape of the trawl gear in function of its speed , for the laid-down line length l = 200 m

Diagram. 8. Simplified shape of the trawl gear in function of its speed , for the laid-down line length l = 300 m

5. FINAL COMMENTS
The elaborated, relatively simple model of the trawl gear as well as calculation software make it possible to determine the gravity centre motion trajectory of the trawl gear immersed in sea water , depending on the main operational parameters such as: ship speed, speed of hauling or laying-down the line by trawl winches , length of the laid-down line as well as amplitude and period of ship stern displacements.

Figure 5. Trawl board and trawl lines to give out from boat.

The model makes it possible also to determine position of the trawl gear main parts as well as divergence (width) of the trawl mouth for definite values of the main design and operational parameters of the gear. This is as much important that the so obtained data can be used during performance of fishing. By appropriate selecting the operational parameters a current position of the trawl and its mouth width can be adjusted to a localized fish shoal , moreover any danger of catching the trawl on underwater obstacles can be avoided this way.

REFERENCES
Bessomneau J.S., Marichel D.: Study of the dynamics of submerged supple nets (application to trawls). Ocean Engineering, 25, 1998 Bielaski J.: The analysis of the system towing rope with long underwater object. Hydroacoustics, 9, Gdask 2006 Brewek D., Eayrs S., Mounsey R., You-Gan W.: Assessment of an environmentally friendly, semipelagic fish trawl. Fisheries Research, 1996 Chun-Woo L., Ju-Hee L., Bong Jin Ch., Hyun-Young K.: Physical modeling for underwater flexible systems dynamic simulation. Ocean Engineering. 32, 2005 Erinfolami L. A.: Investigations of interaction between ship power system and fishing trawl gear ( in Polish). Doctor thesis, Gdask University of Technology, Gdask, 2000 Huang S.: Dynamic analysis of three-dimensional marine cables. Ocean Eng., 21, 1994 Moderhak W.: A calculation method of total hydrodynamic drag of the pelagic trawl net ( in Polish). Doctor thesis, Gdask University of Technology, Gdask 1991 Piechna J. R.: Programming in the language Fortran 90 and 95 ( in Polish). Publishing House of Warsaw University of Technology (Oficyna Wydawnicza Politechniki Warszawskiej), Warszawa 2000 Pogaski Z.: Prospects for Baltic Sea fishery ( in Polish). Nautologia, 2, Gdynia 1998 Wu J., Chwang A. T.: A hydrodynamic model of a two-part underwater towed system. Ocean Engineering , 27, 2000 Wu J., Chwang A. T.: Experimental investigation on a two-part underwater towed system. Ocean Engineering, 28, 2001

CONTACT DETAILS
Czesaw Dymarski, Technical University of Gdask, Gdask, Poland cpdymars@pg.gda.pl

Problem of water flow on deck of small vessel

Monika Warmowska, Jan Jankowski Polski Rejestr Statkw S.A., Gdask, Poland

Green water moving on deck of a small vessel affects its motion and can contribute to the vessels capsizing. The phenomenon is nonlinear and its description using mathematical differential problems continues to be a challenge. The paper presents the comparison study of two simplified methods describing the water moving on deck. The first model is used in RoRo ferries damage stability calculations. The model is based on an assumption that free surface of water over the deck is horizontal. The dynamic pressure caused by water on deck depends on the vessels acceleration and changing height of water. The shallow water model is the second method. This method is used in simulation of shallow water motion along the deck of a small vessel. In this method the vertical acceleration of water is neglected and the horizontal components of waters velocity do not depend on the vertical coordinate. The paper presents verification possibilities of applying these methods in practice.

Keywords: method of shallow water flow, vessel capsizing, water on deck, irregular waves, vessel motion

1. INTRODUCTION Figures reported by the International Marine Organization itself show that the annual loss of life on worlds fishing vessels accounts for the loss of a huge number of human lives every year and that the safety of small vessels is a problem. One of the important reason for capsizing of small vessels is water moving on deck. Correct modelling of hydrodynamic forces imposed by moving water on deck is a very complex and difficult problem if continuously changing amount of water on deck is taken into consideration. A detailed description of this problem is presented by Belenky (2002). Dilligham (1981) presents a formulae describing water flow through scrubbers and over bulwarks. The mass of flowing water depends on the size and shape of openings, the height of sea wave and of the water height above the opening. Numerical solution of 2D problem is obtained by Dillingham (1981). The equation used is valid for a level ship at rest. The author applies the random choice method for solving hyperbolic equations. The three-dimensional flow is described by Dillingham and Falzarano (1986) who transform equation to a coordinate system coupled to the ships centre of gravity. Panatazapoulas (1988) presents a 3D equation of shallow water motion on deck of a ship moving in waves with yow equal to zero. His method is further developed by Huang and Hsiung (1997) who apply the flux differential splitting method to solve the non-linear threedimensional problem describing water flow on deck. The forces and moments of water moving on deck are added to equations of ship motion. Jankowski & Laskowski (2006) applied a simplified approach used in Ro-Ro ferries damage stability calculations in their computer programs enabling simulation of ship motion in irregular waves (Jankowski, 2007), as the first phase of modelling the water-on-deck effects. They added additional pressure acting on the deck to the model used in Ro-Ro ferries, caused by the change of water amount on the deck ( Buchner, 2002). In the second phase the problem of shallow water flow was used to model the water motion on deck, which include horizontal relative velocity of water (Zienkiewicz, 2005). The model was verified (Warmowska, 2008) testing cases of water motion: with constant water mass inside tank moving with constant acceleration, with constant water mass inside tank moving in 3D, trapping on non moving open deck.

The results were compare with results obtained by other authors and in other experiments (Huang Z.-J., Hsiung C., 1997). The paper presents the comparison of these two methods for the given ship motion in time.

2 EQUATION OF VESSEL MOTION WITH WATER ON DECK IN IRREGULAR WAVES The simulation of vessel motions in waves is based on numerical solutions of non-linear equations of motion (non-linear model). The hydrodynamic forces and moments defining the equations are determined in each time step. The accuracy of the simulation depends on the accuracy of calculating the hydrodynamic forces and moments due to waves. It is assumed that the hydrodynamic forces acting on the vessel can be split into FroudeKrylov forces, diffraction and radiation forces as well as other forces, such as those induced by water on deck, rudder forces and non linear damping. The Froude-Krylov forces are obtained by integrating the pressure caused by irregular waves undisturbed by the presence of the ship over the actual wetted ship surface. The diffraction forces are determined as a superposition of diffraction forces caused by the harmonic components of the irregular wave. It is assumed that the ship diffracting the waves is in its mean position. This is possible under the assumption that the diffraction phenomenon is described by a linear hydrodynamic problem. The variables of diffraction function are separated into space and time variables with the space factor of the function being the solution of the hydrodynamic problem and the known time factor. Such an approach significantly simplifies calculations because bulky calculations can be performed at the beginning of the simulations and the ready solutions can be applied for determining the diffraction forces during the simulation. The radiation forces are determined by added masses for infinite frequency and by the socalled memory functions (given in the form of convolution). The memory functions take into account the disturbance of water, caused by the preceding ship motions, affecting the motion of the ship in the time instant in which the simulation is calculated. The volume of water on deck, varying in time, depends on the difference in heights between the wave surface and the following edges: the upper edge of the bulwark, and the lower edges of openings in the bulwark.

The equations (1) of ship motion in irregular waves are written in the non-inertial reference system. The system Q, with axis QX directed to the ship bow, QY directed to the ship port side and QZ directed upward, is fixed to the ship in the centre of its mass and the equations of ship motion assume the following form (Jankowski, 2007):

m[V Q (t ) + L(t ) +

(t ) v Q (t )] = FW (t ) + FD (t ) + FR (t ) + Fd (t ) + FA (t ) + D 1G ,

(1)

(t ) L(t ) = M QW (t ) + M QD (t ) + M QR (t ) + M Qd (t ) + M QA (t ), R UQ (t ) = v Q (t )

(t ) R UQ (t ),

( (t ), (t ), (t )) T = D 1 (t )

where m is the mass of the vessel, vQ = (vQ1,vQ2,vQ3) is the velocity of the mass centre,

(1, 2, 3) is angular velocity, L = (lQ1, lQ2, lQ3) is the angular momentum, RUQ= (rUQ1, rUQ2, rUQ3) is the position vector of the ship mass centre in relation to the inertial system U, moving with a constant speed equal to the average speed of the vessel, ( , , ) are Eulers angles
representing roll, pitch, yaw, FW, FD, FR and Fd are FroudeKrylov, diffraction, radiation forces and forces caused by water on deck, respectively, G = (0, 0, mg) gravity force,

MQW, MQD, MQR, MQd are their moments in relation to the mass centre, D is the rotation
matrix, and D is the matrix which transforms Euler components of rotational velocity

( , , ) into

. The additional forces and moments such as damping forces or those

generated by the rudder are denoted by FA and MQA. The ways of solving 3D hydrodynamic problems and determining forces appearing in the equation of motion are presented by Jankowski (2007). The non-linear equations of motion (1) are solved numerically (Hamming procedure is applied) according to the method presented by Ralston (1975). The program based on equations presented above and on the numerical methods applied enables to perform simulation of vessel motion with water on deck in irregular waves. The example of simulation history of motion is presented in Figure 1.

40

[m]
0 -5 5
wave [m]

[deg]
0 -40

0 -5 150 200 250 t [s] 300 350

40 [deg] 0 -40
150 200 250 t[s] 300 350

Figure 1. Time history of surface elevation, heave , roll and pitch

3. THE FORCES AND MOMENTS ON DECK

The volume of water on deck, varying in time, depends on the difference in heights between the wave surface and the following edges: the upper edge of the bulwark, and the lower edges of openings in the bulwark. It is assumed that the flow rate of water volume over the bulwark can be calculated as the flow over a weir, whereas the flow through the openings in the bulwark is modelled as a flow through a submerged orifice in a dam. The general formula for the flow rate is:

q = ( signH )cb 2 g 2 H 3

3 2

+d H

1 2

(2)

where c is the correction coefficient for non-stationary flow, established experimentally, b is the width of the orifice or the fragment of bulwark above which the deck is flooded, H is the vertical distance between the wave profile and the free surface on the deck at a point considered (positive if the wave exceeds the water level on the deck), d is the depth of water at the orifice or the instantaneous elevation of wave profile above the deck edge at the orifice. Equation (2) assumes various forms depending on relative water levels inside and outside the deck well and on the position of the opening in the bulwark (Pawowski, 2004). The formula (2) is applied separately for the upper edge of bulwark and the openings in the bulwark.

The forces Fd, and moments MQd, caused by water on deck, are calculated according to the following formula:

Fdi = p d ni dS , i = 1,2,..,6
S

(3)

where S is the wetted surface of the deck, n1, n2, n3 are components of the normal vector n in the considered deck surface S and n4, n5, n6 are the components of Rn. Vector R = (xd, yd,

zd) is the position vector of deck point in non inertial system Q (the components of the
normal vector are determined in the reference system fixed to the vessel in its centre of mass

Q), pd is the pressure in point on the deck.

3.1 Model used in Ro-Ro ferries damage stability calculations

The position of water trapped on deck is determined by the horizontal plane and the actual position of the vessel deck in the given time instant. The dynamics of water caused by the motion of water particles in relation to the deck is neglected. The forces and moments caused by water-on-deck are obtained by integrating the hydrostatic pressure determined by water horizontal plane above the deck in the vessels actual position (for time t=ti, i=1..n). Additionally, the vessels acceleration and the changing heights of the horizontal plane above the deck have been added to the model to better replicate the phenomenon ( Jankowski & Laskowski, 2006). The pressure pd in point on the deck is equal to:

pd =

dh vV + ( g + aV )h dt

(4)

where

vV = d 31v P1 + d 32 v P 2 + d 33 v P 3 , aV = d 31 a P1 + d 32 a P 2 + d 33 a P 3 , v P = vQ +

R QP ,

a P v Q + R QP

d3i, i = 1,2,3, are components of the matrix D and h is the vertical distance from the horizontal plane to the point of the deck in the inertial coordinate system U. Velocity vV and acceleration aV are also determined in the system U. This formula has been derived, basing on the evaluation of Newtons momentum relation for a control volume on deck, in the following way (Buchner, 2002):

Fd 3 =

d ( mvV ) d ( m) dv = vV + V m dt dt dt

(5)

where Fd3 is the force acting on the area S of the deck. The mass in the control volume is equal to: m=hS. Substituting this to equation (5), dividing by S and taking into account gravitational acceleration g, formula (4) is obtained.

3.2 The shallow water model The problem is described in the system Q, fixed to the ship in the centre of mass m. The vessel has six degrees of freedom. The force Fd, with components (fQx, fQy, fQz) in the system

Q, acts on the water particle. The gravity forces, centrifugal forces, Coriolis and tangential
forces contribute to the force Fd. The shallow water problem is solved in four steps (Warmowska, 2008), determining: 1. domain occupied by water, 2. pressure field, 3. horizontal velocities ux , uy, 4. vertical velocity uz.

The motion of the free surface SF is described by the following equations:

dx (t ) = u x (t , x,y,z ), dt dy (t ) = u y (t , x,y,z ), dt dz (t ) = u z (t , x,y,z ). dt

(x(t ),y(t ),z (t )) S F (t )

(6)

Equations (6) are integrated using the Runge-Kutta method. The nodes of the net determining the free surface moving in time are updated in each time step by interpolating the function describing free surface over nodes (xdi, ydi, zdi) of constant Euler grid of deck. It is assumed that the velocity field around the ship is the velocity of the undisturbed ocean wave. In the case of the deck submerged in the wave, the water particle velocity over the deck, occurring in equations (6), is calculated taking into account the velocity field being the average of the deck water velocity field and the wave velocity field. It is assumed that viscosity forces can be neglected. The motion of water is described using Euler equations. The shallow water model approximating the motion of water on vessel deck assumes that the vertical acceleration az can be neglected: a z (t , x,y,z ) 0 . These assumptions
results in following equation: The motion of the free surface SF is described by the following equations:

p (t , x,y,z ) f Qz (t , x,y,z ), z

(x(t ),y (t ),z (t )) (t ).

(7)

Equation (7) determines the pressure in the water domain over the vessel deck. The pressure field on deck pd is obtained integrating equation (7):

p d (t , x,y,z d ) p a +

zd

f Qz t , x,y,s zd + h ( t , x , y , z d )

)ds, (x(t ),y(t ),z d (t )) (t ).

(8)

where pa is the pressure on the free surface SF corresponding to atmospheric pressure.

Horizontal components of the velocity field in the domain are determined from the Euler equations:

du x (t , x,y,z ) = f Qx (t , x, y, z ) 1 p (x,y,z ), x dt du y (t , x,y,z ) = f Qy (t , x, y, z ) 1 p (x,y,z ), y dt

(x(t ),y(t ),z (t )) ,

(9)

for pressure determined by formula (8). The vertical component uz of the velocity field in the domain is determined from the equation of mass conservation. Additionally, in the shallow water model it is assumed that horizontal velocities ux and uy do not depend on the vertical coordinate z. Basing on this assumption, the equation determining vertical component uz takes the form:

u y u u z (t , x, y, z ) = x (t , x, y ) (t , x, y ) + q ( z z deck ) x y

(10)

where q represents the changes of waters mass on deck.

4. COMPARISON OF THE RESULTS OBTAINED BY THE MODELS PRESENTED

The forces Fd = (Fd1,Fd2,Fd3) and moments MQd = (Fd4,Fd5,Fd6) caused by water on deck are calculated with the use of formula (3) where the pressure pd is determined by: formula (4) for the model used in Ro-Ro ferries damage stability calculations; formula (8) for the shallow water model. The comparison was made in the following way: vessel motion in waves and forces generated on deck, affecting the vessel motion, is determined with the use of Ro-Ro ferry model, and

 forces generated on deck with the use of the shallow water model are calculated for the same vessel motion. It means that forces generated on the vessel deck were calculated with the use of two different models for the same vessel motion making the comparison of the forces possible. The irregular wave, determined by significant wave height Hs=6m and mean period Tz=8s, followed the vessel moving with the forward speed u=6m/s. The angle between the vector of forward vessel velocity and the wave vector was 30 degree. Figure 2 presents force Fd3 generated by water on deck, which increases the vessel draught, and Figure 6 the rolling moment Fd4, responsible for vessel capsizing.

Fd3 [kN] 200 0 -200 -400 -600 220 240

shallow water method simplified method

260 t [s]

280

Figure 2. Time history of vertical force Fd3 (increasing the draught)

The comparison of the forces determined by both methods shows the following: in the period [240s, 248s) force Fd3 matches quite well for both methods it is the case of the inflow of wave water on deck (Figure 3), when the vessel is in a relatively calm wave trough and the free surface on deck is almost horizontal plane (therefore both methods match well);

Figure 3. Inflow of wave water on deck

 in the period [248s, 260s) the water surface on the deck is still below the wave surface (Figure 4); the shallow water method models water motion on deck and the changes of water level on the deck are not as quick as in the simplified method as in this method the volume of water on deck depends only on the difference between the wave surface and deck water surface and does not depend on the velocity field on the deck;

Figure 4. The relation between wave surface and water surface on deck at t=256s

in the period [260s, 265s] the water surface drops immediately in the simplified method when the vessel bow climbs upwards while in the shallow water method the velocity field in the water on deck slows down the process (Figure 5)

Figure 5. Outflow of water from deck (t=262s)

 in the period considered the rolling moment matches very well for both methods (Figure 6); this is probably because the following wave is considered which, for this conditions, does not generate significant water motion across the deck.

Fd4 [kNm] 800 400 0 -400 -800 220

shallow water method simplified method

240 t [s]

260

280

Figure 6. Time history of rolling moment Fd4

5. CONCLUSIONS

The study covered small vessel motion with water on deck (taking also into account the phenomenon of deck in water) in significant irregular waves. In such waves the vessel is positioned, according to good seamanship practice, in the head or following sea. The relatively rapid outflow of water from the deck due to the openings in the bulwark, means that the phenomenon of water flow on deck differs from that of sloshing in a tank as the water motion is not restricted to rebound between the bulwarks as in the tank between tank walls and the free surface of the water does not deform significantly as in the sloshing phenomenon. Maybe, this results from the observed fact that the simplified method (assuming the free surface of the water on deck to be horizontal plane) is used to model the water on deck. However, the study shows that the velocity field in the water on deck, generated by water inflow and outflow on and off the vessel deck and the vessel motion in waves, influences the forces generated on the deck. In the next step, studies will focus on the beam sea (wave form vessel side) for lower wave height and rolling moment.

6. REFERENCES

Belenky V., Luit D., Weems K., Shin Y.-S., (2002), Nonlinear ship roll simulation with water-on-deck, 6th International Ship Stability Workshop, Webb Institute, NewYork. Buchner B., (2002), Green water on ship type offshore structures, PhD Thesis, Delft University of Technology. Dillingham J.T., (1981), Motion studies of a vessel with water on deck, Marine Technology, Vol. 18, No. 1. Dillingham J.T., Falzarano J.M., (1986), Three-dimensional numerical simulation of green water on deck, 3rd International Conference on the Stability of Ship and Ocean Vehicles, STAB86, Gdask, Poland. Huang Z.-J., Hsiung C. C., (1997), Nonlinear shallow-water flow on deck coupled with ship motion, Proceedings of the Twenty First Symposium of Naval Hydrodynamics, National Academy Press, Washington, D.C. Jankowski J., (2007), Statek wobec dziaania fali, Raport Techniczny Nr 52, PRS, Gdask. Jankowski J., Laskowski A., (2006), Capsizing of small vessel due to waves and water trapped on deck, Proceedings of the 9th International Conference STAB 2006, Brasil. Pantazopoulos M.S., (1988), Three-dimensional sloshing of water on deck, Marine Technology, Vol. 25, No. 4. Pawowski M., (2004), Subdivision and Damage Stability of Ship, Euro-MTEC series, pp.217-220. Ralston A., (1975), First course in numerical analysis, PWN, Warsaw. Warmowska M., (2007), Problem of water flow on deck, Archives of civil and mechanical engineering, Wrocaw, vol. VII, No. 4. Zienkiewicz O., Taylor R.L., Nithiarasu P., (2005), The Finite Element Method for Fluid Dynamics, Elsevier.

CONTACT DETAILS

Monika Warmowska, Jan Jankowski, Polski Rejestr Statkw S.A., Al. Gen. Jzefa Hallera 126, 80-177 Gdask, Polska m.warmowska@prs.pl j.jankowski@prs.pl

Development of Future Generation Stability Criteria

Lech Kobyliski1, 2
1

Foundation for Safety of Navigation and Environment Protection

2

Polish Academy of Sciences, Gdansk, Poland

Recently International Maritime Organization adopted revised intact stability requirements that are included in the Intact Stability Code 2008. Part A of the Code is now compulsory via reference in the SOLAS Convention. In order to further enhance safety against stability failure the Marine Safety Committee of IMO decided to develop future generation stability criteria intended to be applied to non-conventional ships and taking into account failure modes not sufficiently covered by the present Code. The work on those criteria is now well advanced. The author presents critical views re current proposals and proposes to apply riskbased approach as an alternative to the criteria based on the present framework for development of future generation stability criteria.

Keywords: ship stability, stability regulations, risk analysis

1. INTRODUCTION Recently the Sub-committee SLF of IMO adopted International Intact Stability Code, 2008, that is revised version of the previous editions of the Code. This Code enters into force on 1st July 2010. With the development of the revised Code (2008 IS Code) the first step of the programme of improving stability criteria with which the SLF Sub-committee of IMO was charged in 1995 was completed. The most important change in comparison with the original version of the Code adopted in 1993 is that part A of the Code incorporating basic criteria is made compulsory via reference in the SOLAS Convention. The stability criteria as included in the revised Code are virtually the same as in the original IMO resolution A.167(ES.IV) adopted in 1968 (statistical criteria) and in resolution A.562(14) adopted in 1985 (severe wind and rolling criterion) with small amendments and some relaxations. Regarding, however, weather criterion the Code provides possibility of using an alternative method of calculation of wind force based on model tests. From the point of view of ship safety this is however, not the final solution. From time to time, stability casualties happen in spite of the fact that the particular ship meets all existing IMO criteria. The existing criteria may also be not applicable to some types of modern ships incorporating novel design features especially because original criteria as in resolution A.167(ES.IV) developed more than forty years ago were based on casualty statistics that included mainly vessels under 100m in length. With many modern ships there is no previous experience in relation to safety and stability and satisfying existing criteria may not assure required level of safety. Because of this, Maritime Safety Committee of IMO has recently included in its work programme the development of so called new generation criteria. (IMO 2008). Work on the development of those criteria is now well advanced.

2. RECENT DECISIONS OF THE IMO SLF SUBCOMMITTEE At its 51st session in 2008 the SLF Sub-committee agreed upon the framework for the new generation stability criteria (IMO 2008a). The framework includes definitions of stability failures (total or partial), definitions of types criteria (deterministic or probabilistic,

parametric or performance based) and the concept of vulnerability criteria. According to the decision of the Sub-Committee, new generation stability criteria are intended basically for non-conventional ships and should be used as alternative or supplement to existing criteria.

The new generation stability criteria should take into consideration primarily three modes of failure: restoring arm variation problems on wave crest such as parametric excitation and pure loss of stability stability under dead ship condition, and manoeuvring related problems in waves such as surf riding and broaching-to

The above short description of the task was little more elaborated in the Framework for new generation stability criteria and associated terminology included in the document (IMO 2008a) with further explanation in the paper by Belenky, de Kat and Umeda (2008). The framework, apart from inclusion of definitions of basic terms, included also proposal to apply vulnerability criteria that would check for the susceptibility of the ship to various modes of stability failure and that should be based on simplified models, simple mathematical formulations, analytical solutions or statistical data. Vulnerability criteria are intended to distinguish between conventional and non-conventional ships. General procedure of safety assessment in the future would be as shown in the Table 1.

Table 1. Methods of safety assurance

SHIP TYPES Conventional Non-conventional

Method of stability safety assessment Prescriptive criteria as in the IS Code 2008 New generation stability criteria

3. CONCEPT OF DEVELOPMENT AND APPLICATION OF NEW GENERATION STABILITY CRITERIA The working group on intact stability created by the SLF Sub-Committee IMO discussed in depth the possibilities of development of future generation intact stability criteria and came to the conclusion that they should follow the scheme that is shown in the Figure1. In this scheme three levels of application of the criteria are proposed (IMO 2010).

Figure 1. Proposed structure of the new generation stability criteria.

Level one and level two are vulnerability criteria, the most simple ones in the level one, the little more elaborated in the level two. If the particular ship would pass level one and level two criteria, then requirements of IS Code 2008 would apply. If it will not pass those criteria, then level three would apply that includes direct assessment procedure. The working group at the 52nd session of the Sub-Committee developed also detailed specification for the particular levels of the proposed structure as shown in the table 2 (IMO 2010a). Vulnerability criteria levels 1 and 2 will be developed with the purpose of assessing stability failures not covered by existing criteria and also providing justification for the application of direct stability assessment. Understanding that direct stability assessment has higher level of complexity, it is expected to be applied only to ships deemed vulnerable for a particular failure mode (as determined by the vulnerability criteria level 2). The standard for the level 2 vulnerability criteria will be chosen to provide sufficient justification for the application of direct stability assessment.

Table 2. Specification of the proposed levels of the future generation stability criteria

Level

Complexity Safety margin Higher VulnerabilityFormulae or simple procedure based on Low level 1 geometry/ hydrostatics load condition standards and hydrostatic parameters are needed Description

Objective

VulnerabilitySimplified physics-based calculations Moderate level 2 with reduced computational efforts and straightforward application following suitable guidelines

Direct These criteria should be based on the High stability best state of art concepts available. assessment The following is envisioned as a minimum requirement to direct stability assessment methods to ensure physical robustness. Time-domain numerical simulation with hybrid method and probability theory, as appropriate, should be considered. The hybrid method includes potential flow plus empirical viscous models. Specifically, rigid body-nonlinear dynamics model with undisturbed wave pressure (Froude-Kriloff assumption). Specified formulations for added mass/wave damping/diffraction, externally specified coefficients for viscous/hydrodynamic lift components of roll damping and manoeuvring, and propulsion force, external environment actions should be included, as appropriate. Suitable guidelines and procedures (e.g. wave scatter diagram, ship operation conditions, etc) should be clearly stated. Assessment is expected to be made using a probabilistic measure to evaluate safety level. N/A Operational Ship-specific operational guidance, if guidance applicable, should be based on the output of the direct stability assessment procedure and on the analysis of stability failures

To determine if the ship shall be considered unconventional for a specific mode of stability failure Moderate To confirm vulnerability to a specific mode of stability failure and justify the application of direct stability assessment for this mode To evaluate safety Lower standards level and prepare information for the can be applied development of ship-specific operational guidance

N/A

To assist with safe operation of a vulnerable ship

At the International DSCSD Workshop several proposals concerning different criteria for level 1 and 2 and for direct assessment for four modes of stability failure were presented by different countries (Kobyliski 2009). They were reviewed at 52nd session of the SLF SubCommittee and the results are summarised in the table 3. Particulars of these proposals may be found in the proceedings of the DSCSD Workshop.

Table 3. Summary of the methodologies for different stability failure modes

Stability failure mode Pure loss stability Parametric roll Surf riding/broaching Dead ship condition

Leve1
of

Level2 United States United States Japan Italy, Japan

Direct assessment Operational guidance Germany Germany Germany Japan Italy, Japan Germany Germany

Japan Italy, Japan

4. CRITICAL COMMENTS REGARDING LATEST PROPOSALS CONCERNING FUTURE GENERATION STABILITY CRITERIA The merit of the proposal consist of taking into account first of all hazards posed by some phenomena occurring when the ship is sailing amongst the waves, as pointed out above. Those dangerous situations were well recognized 25 years ago, when the SLF Subcommittee at its 29th and following sessions recognized already that the intact stability criteria as included in the Resolutions A.167(ES.IV) and A.168(ES.IV) statistical criteria, and also in the Resolution A.562(14) weather criterion, were not adequate and considered possibility of developing criteria based on probability of capsizing for certain dangerous situations not covered by the criteria, namely: 10 ship sailing in following seas, 20 ship positioned in beam sea in gusty wind (dead ship condition), and 30 ship sailing in quartering seas in conjunction with broaching. These were virtually the same situations as considered recently. At that time this concept was discussed in several papers, e. g by Boroday and Rakhmanin (1975), Kobyliski (1975), Takaishi (1982) Cleary (1975, Dorin et al (1975) and formally proposed to IMO by Poland (IMO 1978). The problem at that time was apparently too difficult to solve and the idea was ultimately abandoned. The most important issue in the development of future generation stability criteria is proper choice of modes of capsizing. Capsizing modes or scenarios should take into account

conclusions taken from

statistics and from records of casualties. The other sources of

information are model tests and interesting conclusions regarding possible way of capsizing could be drawn from observation of model behavior in waves. In order to develop criteria the set of capsizing scenarios should be considered. It seems, however, that the number of possible scenarios leading to capsizing is extremely large. When using simple physical models in many cases analytical solutions may be adequate and in consequence simple deterministic criteria are developed. However simple scenarios are rare in reality and analysis of historical data on casualties almost always revealed that stability failure is the result of sequence of events that may be attributed to different causes and where human factor usually played predominant role. In order to assure safety against capsizing and to identify possible ways of stability failure system approach should be adopted. This was pointed out by the author as early as in 1984 and also later in several papers, e.g. (Kobylinski 1984, 2007). Ship stability system is rather complicated. However, in most cases it could be considered as consisting of four basic elements: ship, environment, cargo and operation (Kastner 1986). Those elements are

strongly interconnected. Analysis of stability casualties reveals that the causes of casualty may be attributed to (Cleary 1975, Erickson et al 1997): functional aspects resulting from reliability characteristics of the technical system, therefore stability characteristics of the ship operational aspects resulting from action of the personnel handling the system,

therefore crew members but also ship management, port authorities, marine administration and owners company organisation external causes resulting from factors independent from designers, builders and operators of the technical system therefore ship environment and climatology cargo related aspect resulting from characteristics of cargo and its way of transporting.

In order to achieve sufficient level of safety with respect of stability, all elements creating stability system have to be taken into account. Taking into account the fact, that less than 20% of all casualties are caused by faulty or bad design of the ship, the existing safety requirements that refer mainly to design features of the ship cannot ensure sufficient level of safety, in particular with regard to ships having novel design features.

The Framework for development of new generation stability criteria agreed by the SLF Subcommittee (IMO 2008a) does not include any hints as to the form of the future criteria nor as to the tools that should be used. The general layout of the document leads however to the conclusion that the intention is to develop future criteria as prescriptive criteria, however performance based, and basically design oriented. The weak point of the proposed approach is that only three dangerous situations were selected for future consideration. Statistical data of stability casualties do not indicate, however, that those situations are the most common cause of stability accidents In the opinion of the author the approach proposed by the working group restricts the purpose of the future criteria and requirements especially with the view of goal oriented approach. There certainly is merit in consideration of the hazardous situations mentioned above, but the possibility to develop prescriptive requirements and formulae that could be used as criteria taking account the above phenomena is doubtful. This conservative approach seems to be inadequate. It is true, that it is very difficult in the majority of stability failures to discover a single cause of casualty. Usually an accident is a consequence of the chain of events where human failure plays predominant role. Very seldom it is possible to reveal that accident was caused solely by bad design, especially that ships should satisfy stability requirements in force. The analysis of causes of 364 stability casualties collated from various sources (Kobyliski 2008) allowed to draw some general conclusions revealing that in the great majority of cases casualties (about 80%) human and organizational errors are responsible for the accident that is usually the result of a sequence of events that involve environmental conditions, ship loading condition, ship handling aspects and human factor in general and usually it is associated with some other factors such as shifting of cargo the most frequent event - or with water inrush Most accidents took place in rough sea, although forces of the sea were not always the primary event. Several casualties happened in comparatively calm sea. Analysis of causes of stability failures reveals also that design features of the ship are responsible for a small percentage of casualties In each hazardous situation different capsizing scenarios may develop. Capsizing scenario is the sequence of events leading ultimately to stability failure, either total that means capsizing, or partial, that means large list with loss of power, loss of containers or disabling the ship. Many scenarios of capsizing were identified, mainly based on detailed description of

accidents and also on observations of the capsizing of models tested in towing tanks or in open waters in waves. Experts opinions contributed greatly to the creative part of identification of scenarios where operational aspects and human performance played important part. Block diagram in Figure 2 shows hazardous situations identified and possible capsizing scenarios only due to forces of the sea, taken from various sources (Kobylinski 2003. de Kat et al., 1994) see also Kobylinski (2007b). This should be considered as an example; in reality certainly the number of scenarios may be larger. Quite often attention of seafahrers is drawn to certain mode of failure after casualty happened. This was the case for example with the accident of container vessel C11 due to parametric resonance in head seas (France et al 2001) , although even earlier parametric resonance in head seas was observed (Burcher 1990). More recently situation of meting freak wave was added to hazardous situations requiring consideration because of casualties reported (e.g. Buckley 1997; Faulkner&Williams 1996; Pierson 1993). Also breaking waves that caused capsizing of the fishery research vessel HELLAND HANSEN (Dahle&Kjaerland 1999) were brought to the attention. In fact, however, parametric resonance, the mode most seriously considered within the working group is not the most important and most often mode of stability failure. There is no evidence of capsizing of a container ship due to parametric resonance, although there are recorded many losses of containers due to excessive accelerations in following of head seas. This might be due to excessive rather than inadequate stability and in fact those situations might be avoided by proper operation of the ship in question (IMO 2010b). It should be stressed that vulnerability criteria for this mode of failure as well as for pure loss of stability on wave crest were not yet developed and in fact the discussions did show that it might be impossible to develop them, because almost every ship may be vulnerable to such phenomena in certain environmental conditions and the only way to avoid such situations is by developing suitable operational guidance. In the view of the author when considering future generation stability criteria all modes of capsizing should be taken into account and the attention of the law makers should be concentrated on those modes of capsizing to which greatest probability of occurrence could be attached. At present there was no attempt to attach probabilities to vulnerability to four modes of failure as shown in the table 3 and the vulnerability criteria are intended to say yes or no which seems to highly inadequate.

Figure 2. Schematic presentation of hazardous situations and possible scenarios of capsizing

5. ADVOCATING RISK BASED APPROACH Many difficulties with the development of the future generation stability criteria could be avoided if risk based approach would be applied. Risk analysis is considered as an alternative to the prescriptive criteria. The basic dichotomy exists between prescriptive and risk-based requirements, as shown in Figure 3.

Figure 3 . Prescriptive versus risk-based criteria

Risk based requirements have many advantages in comparison with prescriptive criteria, Prescriptive criteria are traditional, usually based on deterministic calculations, they are easy to understand and apply. On the other hand they are bounding the designer and they do not allow to apply non-conventional design solutions. Risk- based approach could be described as goal-oriented performance based approach utilizing, as a rule, probabilistic calculations. Application of risk-based approach to IMO rule-making process was considered firstly in the fall of 20th century by the Maritime Safety Committee and finally recommended in the form of Formal Safety Assessment (FSA) (IMO 2002). Author advocated application of safety assessment based on risk approach to stability and application of FSA to stability in several papers, e.g. (Kobyliski 2003, 2004). The response to those proposals was not very

encouraging, however, at the beginning, but slowly the ideas found their way, mainly because in other areas of shipbuilding technology the need to apply risk analysis was well recognized (e.g. SAFEDOR project, Papanicolaou 2009). Recently IMO took the decision to use risk approach in new approval process requirements to be developed (Anonym 2009). Actually Framework for development of new generation intact stability criteria as adopted, is partially based on risk analysis, not mentioning this term however, although not in consequential way. Vulnerability criteria are actually nothing else as hazard identification. Direct assessment methods include probability of failure calculation. What is missing are hazard probabilities and risk assessment procedure. Also, safety level is not defined in the Framework and, if one understands safety level as risk, there no indications in the Framework how the risk should be evaluated and accepted.

In the opinion of the author, would the risk-based approach be used consequentially, the task of developing criteria for different modes of stability failure would be much easier. For example, vulnerability criteria level 1 and 2 for parametric resonance could be easily formulated, because within risk-based approach different tools could be employed, as e.g. expert opinions, historical data, model tests and not necessarily only mathematical algorithms or formulae. It would be also possible to handle complex stability failure modes including human factor and operational faults. A very good comparison of features of prescriptive approach and riskbased approach was presented in the paper by Luhmann and Maccari (2009) see Table 4.

Table 4. Comparison of features of prescriptive criteria and risk-based approach

Design improvement

PRESCRIPTIVE progressive

Compliance with regulations full Direct calculations Limited amount

RISK-BASED Significant evolution, new prototypes substantial Large use Innovative, new certification process Driven by risk-based requirements Cost-benefit analysis to support innovative solutions Direct, complex time consuming

Equipment and suppliers Innovation Cost Plan approval process

Well known Limited Driven by new product availability Savings based upon economy & minimum requirements standard

It seems that one reason why people are rather hesitant to use risk-based approach is that they think this procedure is to complicated. In fact it is not so. The basic block diagram for risk analysis is shown in fig. 4. The author did show in several papers how this procedure could be applied to stability problems, and in this context reference is made to the following papers (Kobyliski 2003, 2004, 2005a, 2007, 2007b and 2008).

Fig.4. Block diagram for the risk-based approach

6. CONCLUSIONS The decision of IMO to start developing future generation stability criteria is a very important one from the point of view of enhancing safety at sea. However, the adopted framework for development of those criteria raises some important questions and critical comments. The author sees many advantages using risk-based approach to stability problems instead of the approach proposed as shown in the paper. In the opinion of the author, at present, risk-based approach is used successfully in many fields of technology and also in shipbuilding technology and its advantages are well substantiated.

REFERENCES Anonym (2009): Risk-based rules to find form at IMO. The Naval Architect, March Belenky V., de Kat, J.O., Umeda N.(2008): Toward Performance-Based Criteria for intact stability. Marine Technology and SNAME News, Vol.45, No.2 Boroday I.K., Rakhmanin N.N. (1975): State of art on studies on capsizing of an intact ship in stormy weather conditions. Proceedings of the 14th ITTC Conference, Ottawa. Buckley W.H. (1997): Critical capsizing conditions in survivability seaways. , Proceedings, of the International. Conference: Design and Operation for Abnormal Conditions, RINA, Glasgow

Burcher R.K. (1990): Experiments into the capsize of ships in head seas. Proceedings of the 4th International STAB Conference, Naples Cleary W.A. (1975): Marine stability criteria., Proceedings of the 1st International STAB Conference, Glasgow Dahle E.A., Kjaerland O. (1979): The capsizing of M/S HELLAND HANSEN . Transactions RINA de Kat O., Brouwer R., McTaggart K., Thomas, W.L. (1994): Intact ship survivability in extreme waves: new criteria)from research and navies perspective. Proceedings of the 5th International STAB Conference, Melbourne, Florida Dorin V.S., Nikolayev E.P., Rakhmanin N.N. (1975): On dangerous situations fraught with capsizing. Proceedings of the 1st International STAB Conference, Glasgow Faulkner D., Williams R.A. (1966): Critical survival conditions for ship design. Design and Operation for Abnormal

Proceedings, of the International. Conference: Conditions, RINA, Glasgow

France, W.N., Levadou, M., Treakle, T.W., Paulling, J.R., Michel, R.K., Moore, C., 2001: An investigation of head sea parametric rolling and its influence on container lashing systems. SNAME Annual meeting. IMO (1978): Intact stability. General philosophy for ships of all types. Submitted by Poland. Doc. STAB XXIII/4 IMO (2002): Guidelines for Formal Safety Assessment (FSA) for use in the IMO rule-making process. Doc. MSC/Circ.1023; MEPC/Circ.392. IMO (2008) Report to the Maritime Safety Committee. Sub-committee on stability, and load lines and on fishing vessel safety, Doc. SLF 51/17 IMO (2008a) Revision of the Intact Stability Code, Report of the Working Group (part 1), Doc SLF 51/WP.2 IMO (2010): Two-layered vulnerability criteria and direct assessment of ship stability. Submitted by Poland. Doc SLF 52/3/3 IMO (2010a): Development of the New generation stability criteria. Report of the Working Group (part1).Doc. SLF 52/WP.1.

IMO (2010b): Proposal with regard to the scope of new generation criteria. Submitted by Germany. Doc. SLF 52/3/5 Kastner S. (1986): Operational stability of ships and safe transport of cargo. Proceedings of the 3rd International STAB Conference, Gdask Kobyliski L. (1975): Rational stability criteri and probability of capsizing. Proceedings of the 1st International STAB Conference, Glasgow Kobyliski L. (1984): Philosophishe und Hydrodynamische Probleme der Internationalen Kenterkriterien von Schiffen. International. Schiffstechnische Symposium, Rostock Kobyliski L. (2003): Capsizing scenarios and hazard identification. Proceedings of the 8th International STAB Conference, Madrid. Kobyliski L. (2004): Application of the FSA methodology to intact stability criteria. Marine Technology Transactions, Vol 15, pp 319-329 Kobyliski L.: (2005): Performance oriented criteria. . Proceedings of the HYDMAN Conference, Ostrda Kobyliski L.(2007): Stability of ships: risk assessment due hazards created by forces of the sea. Archives of Civil and Mechanical; Engineering. Vol.VIII, No.1. Wrocaw Kobyliski L.(2007a): System and risk approach to ship safety, with special emphasis of stability. Proceedings of the International HYDRONAV Conference, Polanica Zdrj Kobyliski L. et al. (2008): Final report of the research project concerning stability. Ilawa 2008 Kobyliski L. (editor) (2009): International Workshop on Dynamic Stability Considerations in Ship Design, Proceedings. Iawa 2009 Luhman H., Maccari A. (2009): Shipyard wiev on risk-based design. The Naval Architect Papanicolaou A. D. editor (2009): Risk-based ship design. Methods tools and application. Springer Pierson W.J. (1993): Ship stability in heavy weather: the real situation and models thereof. U.S. Coast Guard Vessel Stability Symposium, New London Takaishi Y.(1982): Consideration of dangerous situation leading to capsize of ships in waves. Proceedings oft the 2nd International STAB Conference, Tokyo International

CONTACT DETAILS Lech Kobyliski, lechk@portilawa.com

DRIFT PATH PREDICTION OF DAMAGED TANKERS Lech Kobyliski, Jacek Nowicki, Jerzy Oleksy & Tomasz Jaworski Foundation for Safety of Navigation and Environment Protection. Ilawa, Poland

Tanker casualties cause often serious environment pollution with disastrous effects. In order to prohibit or mitigate such disasters it is essential to organize effective salvage operation. Effective salvage operation requires that salvage ships reach disabled ship as early as possible. Damaged ship is drifting under influence of wind, wave and sea current and it is essential to forecast the drifting path of the ship in order to know the predicted position of the damageg ship and to properly organize salvage operation. Programme of research was installed in the Ilawa Ship Handling Research and Training Centre (SHRT) resulting in the development of the method of prediction of the path of disabled tankers in various damage conditions after casualty, partially flooded or broken down in parts, and of suitable computer code allowing fast calculation of the predicted path. Computer code uses hydrodynamic and wave force coefficients for damaged ship based on model tests performed in the towing tank where the model was in different damaged conditions that may happen in practice. The developed computer code was validated by model tests using large model in four different damage situations in natural wind and waves on the lake. Keywords: tanker casualties, drift path, drift resistance, damaged ship, model tests

1. INTRODUCTION In spite of enhanced international safety rules adopted regarding construction of tankers during the last period, where double hull, segregated ballasts, duplicated rudder gear amongst other measures are required, tanker casualties happen with resulting oil spill, sometimes as large as more than hundreds thousand tons. The most serious tanker casualties that happened during the years 1967 to 2008 are shown in the table 1 ( Moyano 2007). In the last column amount of oil that went to the sea is shown, and largest oil spills polluting the coastline are marked by star. Those casualties had the most serious consequences.
Table 1. Most serious tanker casualties and oil spills from tankers in the years 1967-2004

Month/Year 03/1967 03/1971 08/1974 01/1975 05/1976 02.1977 03.1978 11.1979 07.1979 02.1980 01.1983 08.1983 11.1988 03/1989 12/1989 04/1991 05/1991 04/1992 12/1992 01/1993 02/1996 01/1997 12/1999 11/2002 08/2003 12/2007

Ships name Torrey Canyon* Wafra Metula Jacob Maersk Uruiola Havaian Patriot Amoco Cadiz* Independent* Atlantic Express* Irenes Serenade Assimi Castillo de Bellveder* Odyssey Exxcon Valdez* Khark 5 Haven ABT Summer Katina Aegean Sea Braer Sea Empress* Nakhodka* Erika* Prestige* Tasman Spirit Hebei Spirit

Place Scilly Isles South Africa Magellan Strait Oporto, Portugal Galicia, Spain Hawaii Brittany, France Bosphorus Tobago, West Indies Navarino Bay Oman South Africa Nova Scotia, Alaska Morocco Genoa, Italy Angola Mozambique Galicia, Spain Shetlands Milford Haven Japan Sea Brittany, France Galicia, Spain Karachi South Korea

Oil spill/sunk (t) 119 000 65 000 50 000 88 000 101 000 95 000 223 000 95 000 287 000 100 000 53 000 252 000 132 000 37 000 80 000 144 000 260 000 72 000 74 000 85 000 72 000 6 000 20 000 76 000 28 000 10 500

The most often

causes of tanker disasters are collisions with other ships, explosions,

groundings on the rocks and hull breaking. The first serious tanker casualty that happened in European waters happened in 1967 when the supertanker TORREY CANYON, because of the master fault, hit the Seven Stone rocks near the Scilly Island, the hull of the tanker was ruptured and spilled oil caused the most serious ecological disaster. Several other serious tanker casualties happened in the years 1967

to 1977 as shown in the table 1 and the casualties stirred public opinion and maritime authorities and as a result extraordinary conference was organized by the International Maritime Organization (IMO) in 1978 and Protocol to SOLAS and MARPOL Conventions was adopted that included important amendments to those conventions related to tanker safety (IMO 1978). This conference almost coincided with the most serious disaster of the tanker AMOCO CADIZ in 1978 close to shores of Normandy that was caused by rudder gear failure and resulting drifting of the disabled ship towards the rocks where the hull was eventually split into two parts and more than 200 000 tons of oil transported spilled into the sea. In January 1997 Russian tanker NAKHODKA was broken into two parts in Japan sea in large waves, the bow part of it drifted towards Japan and spilled oil caused the worst pollution of the shoreline in Japans history. In December 1999 tanker ERIKA was broken during stormy weather into two parts near Brest. The salvage company attempted to tow the stern part of the ship towards the safe place, but the attempt failed and ultimately both parts of the ship sank and the spilled oil polluted large area of the shoreline (Fig.1)

Fig.1. Bow part of the tanker ERIKA drifting towards the coast

In November 2002 in Greek owned tanker PRESTIGE one of the tanks started leaking due to hull damage in Atlantic Ocean close to Portugal. Because all ports refuse to allow the leaking tanker to enter, the master was forced to sail farther from the shore. Attempts to tow the disabled ship failed and during the stormy weather the ship was broken into two parts. as

shown in Fig. 2, that ultimately separated and finally foundered. The spilled oil drifting towards the Spanish coast caused the worst disaster in the history of Spain.

Fig. 2. Tanker PRESTIGE broken down drifting in Atlantic Ocean

The consequences of such large oil spills are disastrous for the sea life (birds, animals and fish) for the economy (fishery, tourism, recreation) and cleaning operations costs usually many millions euro. Understanding that such disasters, in spite of whatever prohibitive measures are taken, are unavoidable, people realized that it would be necessary to develop a system to prevent or mitigate consequences caused by spilled oil if such casualty happens. This was done for the first time after the NAKHODKA disaster, where in Japan important project of this type was installed under the leadership of professor Hara. This was reported in several places, e.g. (Hara 2000;Hara 2000a; Hara, Mitaka 2000; Hara et al 2000). Attempts to develop certain tools designed to help the salvage organization to deal with the consequences of similar disasters were undertaken also after the ERICA and PRESTIGE disasters (Dand 18981, 1981a; Craneguy 2000; Hara et al 2004; Hoshino et al 2004); Baltic Sea is a closed sea particularly vulnerable to pollution. If disaster of a large tanker, such as described above, happens in the Baltic, the results would be catastrophic to the countries surrounding Baltic Sea. In the extreme case the whole life in the Baltic might be destroyed and the beaches made unusable for the long time. Fortunately, such disaster did not happened yet, but the risk that it may happen in future is rather high. Some indication what

may happen was shown in December 2009, when the bunker ship ROMANKA lost manoeuvrability, drifted towards the Latvian coast and grounded. The ship was without cargo and on the sandy beach hull was not ruptured. Fortunately nothing serious did happen (Anonym 2009). In the Baltic Sea. 55 000 ships pass the Danish Strait every year, and about 6 000 of them are large or medium size tankers carrying dangerous goods; more than 50 ferry routes cross the Baltic Sea. Collision and grounding may easily happen. Therefore it is extremely important to assess risk of such a situation and to be prepared to arrange appropriate salvage action. 2. RESEARCH PROJECT ON PREDICTION OF DRIFT OF DAMAGED TANKERS Bearing in mind the possibility of a large tanker being damaged because of grounding or collision in the Baltic Sea the programme of research was installed in the Foundation for Safety of Navigation and Environment Protection with the aim to predict path of the drifting damaged vessel under the influence of wind, waves and sea current. This task constitutes one important element of the system of preventing pollution in case of tanker casualty allowing fast intervention of the salvage ships. The main aim of the research project was development computer code for prediction of the path of disabled ship. The ship disabled, partially flooded, broken down and without propulsion is drifting freely under the action of wind, waves and current. Prediction of the path of such ship or its parts is possible if all forces acting on such a ship are known, that is hydrodynamic forces acting on the submerged part of the ship, aerodynamic forces (wind forces) acting on the above water part of the ship, wave forces acting on the ship and current forces. Some categories of the above mentioned forces could be calculated using known methods, however with respect to others there are no available data. This in particular applies to hydrodynamic and wave forces. Some data on hydrodynamic forces acting on the underwater part of the hull for the drifting undamaged ship were published, e.g. (Eda 1972; Inoue et al 1981), but for ships damaged no such data are available. Within the scope of the project hydrodynamic and wave forces for the damage ship in different conditions were estimated on the basis of model tests. The model test were arranged in the towing tank owned by the Technical University of Gdansk. Within the scope of the project model tests using large model in different damage conditions were arranged on the lake in order to study the behaviour of drifting damaged ship when

drifting and to validate the developed computer code for prediction of drift path. Those test are unique, because no such test were performed anywhere yet. As a sample ship for model tests tanker of the common size met in the Baltic Sea was chosen. The dimensions of the chosen ship are shown in the table 2 , together with dimensions of the model used to estimate hydrodynamic and wave forces in the towing tank as well as the dimensions of the large model used in experiments on the lake.
Table 2. Dimensions of the sample ship and its models

SHIP [m] Scale LPP LOA B H T1 T2 1 306.0 324.0 54.9 32.25 15.50 21.50

MODEL 1[m] 1:41.06 7.45 7.89 1.34 0.785 0.377 0.523

MODEL 2[m] 1:150 2.04 2.16 0.366 0.215 0.103 0.143

3. PHYSICAL MODEL OF DRIFTING MOTION OF A DAMAGED TANKER After loss of power vessel starts drifting under action of external forces. In this situation the drifting ship is affected by hydrodynamic force acting on the underwater part of the hull, wind force acting on the above water part of the ship, wave force, inertia force and Coriolis force. If we assume that after casualty happens it speed is equal zero, it starts accelerating and rotating until hydrodynamic forces acting on underwater part of the hull became equal to external forces and the sum of moments of hydrodynamic and external forces is equal zero. If external conditions are stationary then the vessel reaches quasi-static position. It is assumed that gravity force and buoyancy force are in equilibrium state and are not taken into account when considering motion in horizontal plane. In the first phase of drifting the ship drifts and rotates until all forces reach momentarily equilibrium condition. After that, if external conditions are stationary, ship starts drifting with constant speed without rotation. The first phase of drift is comparatively short and as observations of model tests and calculations performed show, for the real ship is counted in minutes rather than hours (Kobylinski et al 2009, US Coast Guard 1998). Therefore this phase has no effect on the prediction of drift in long time. This conclusion was confirmed also by Dand (1981) and Gusiev (1954). Observations taken onboard of real ships show that the drift velocity varies from 2 to 10% of wind speed. Undamaged ships drifts usually with ship side exposed to the wind, however

they do not necessarily drift in the direction of the wind but may be deviating from this direction by 600 to one or the other side (HAZMAT 2000; Smeaton 1981). 4. MATHEMATICAL MODEL FOR PREDICTION OF DRIFT OF DAMAGED SHIP Mathematical model for prediction of drift of damaged ship was developed taking the assumption that in short period of time external conditions do not change and therefore the equations that describe situation where the ship reaches equilibrium state, i.e. where the resulting moment of all forces acting on the ship is equal to zero may, be used. Drift motion in quasi-static situation could be described by three algebraic equations in the earth fixed reference axis as follows (Dand 1981; Gusiev 1954; Kobylinski et al 2009; Smeaton 1981):

YH YW Y A YG 0 N H NW N A 0

X H XW X A X G 0

In the above set of equations XG and YG are the surge and sway force components arising from the effect of the Coriolis acceleration All other force and moments components are functions of several variables. Thus we have: Hydrodynamic components: X H X H U , , Wave force components: Wind force components:

X W X W U , Tz , H s , , X A X A w ,Vw ,

Identical expressions are found for Y and N components. In the above equations: HS and TZ significant height and period of waves, - angle between symmetry plane of the ship and direction of incoming waves, Vw- wind velocity, R angle of wind velocity vector, U drift velocity, drift angle. This is quasi-static situation that happens shortly after the ship starts drifting in stationary conditions of the environment, i.e. wind, waves and current. Because changes of the environmental conditions are rather slow, and their periods are much greater than period of yaw then in the short period of time those conditions may be assumed as stationary (Dand 1981a, Dand & Read 1981).When they change, the ship after short transitional period reaches another stationary situation. Thus the drifting path could be divided in a number stationary situations and in each of them the above equations could be applied (Kobylinski et al 2009). Aerodynamic (wind) forces components are calculated employing standard method used for example in weather criterion when assessing stability. It is assumed also that Coriolis

acceleration is changing linearly with drift velocity: aC 2U sin , where: latitude, angular velocity of earth rotation. Current is taken into account in a way that vector of current velocity is added to the vector of drift due to wind and wave forces. However there is lack of data for hydrodynamic and wave force components for the ship in damage conditions. In order to estimate these components, model tests were arranged in the towing tank owned by the Technical University of Gdansk of the dimensions: length L=30.00m, breadth B = 3.00m and depth h=1.5m. Model of the dimension suitable to the size of the towing tank of the tanker chosen was made in the scale 1:150. Model was divided in several compartments that may be flooded and tested in four different damage states: 1. model not damaged, but partially flooded in intact condition, 2. model with flooded forward tanks, trimmed to the bow, the bow submerged, 3. model broken down in the middle, both parts still holding together, 4. stern part of the model separated, floating vertically. Those conditions of damage were chosen in order to reproduce conditions that were observed during the NAKHODKA, ERIKA and PRESTIGE casualties. For details of these tests reference is made to (Kobylinski et al 2009). In the towing tank model towed by the carriage was free to roll and pitch with drift angles changing within the limits 1800. Longitudinal force FX, transverse force FY, moment MZ, and heel angle were measured and recorded. Test were carried for the basic carriage velocity vm=0.126 m/s and for some chosen drift angles also for the velocities vm2=0.07m/s and vm3=0.126m/s. Tests were performed in calm water in order to estimate hydrodynamic forces and in regular waves of the following parameters: h=0.037m, T=0.5935s and =0.55m in order to estimate wave forces.
1,2 1,0 0,8 1,0 0,6 0,4 0,2

0,8

Fx [N]

0,6

Fy [N]
0,070 0,080 0,090 0,100 0,110 v [m/s] 0,120 0,130 0,140 0,150 0,160

0,0 -0,2

0,4

-0,4 -0,6 -0,8

0,2

0,0 0,060

-1,0 0,060

0,070

0,080

0,090

0,100

0,110 v [m/s]

0,120

0,130

0,140

0,150

0,160

Fig. 3. Typical plot of forces measured in the towing tank

45

90

135

Fig. 4 Model during tests in the towing tank

45

90

135

Parameters of the wave were adopted to correspond to the significant height of maximum waves in Baltic Sea equal to: hS=5.5m and S=82.5m. Measurements of forces and moments

were presented in the coordinates system xyz fixed to the towing carriage. Typical plot of measured quantities is shown in fig. 3 whether the model in the towing tank is shown in fig. 4. Computer code was developed for calculation of the predicted path of drifting ship taking into account the above assumptions. 5. VALIDATION OF THE METHOD. MODEL TESTS OF DRIFT OF DAMAGED SHIP ON THE LAKE JEZIORAK In order to validate the developed computer code for prediction of drift of damaged ship experiments were arranged where large model in different damage conditions was tested on the lake Jeziorak drifting under the influence of wind and waves and current. Those are unique experiments, never before performed anywhere. The experiments were preceded with experiments on lake Silm performed in the autumn 2008 with the purpose: to check reliability of the measuring system to check the practicability of testing method The preliminary experiments were reported by Kobylinski et al (2009). The experiments with damaged ship model were performed in the autumn of 2009 and the large scale model of the chosen type of tanker (Table 2) was tested in windy weather in four different damage conditions shown in fig. 5. Actually, those were the same conditions as tested in the towing tank. -. 2

1 3

Fig. 5.. Model in different damage conditions

During the tests several quantities were recorded:

- trajectory of the model, its heading and velocity, also wind velocity and direction. Those quantities were recorded in the model itself
-

wind velocity and direction at three levels above the water level, wave elevation current velocity and direction. Those quantities were measured and recorded in the fixed position on the lake in the area of experiments. Those data are transmitted to the model and recorded there together with trajectory of the model taken from satellite receiver and electronic compass.

The principle of system of measurements is shown in fig. 6 and 7

Figure 6. Measuring apparatus in the model., 2radiomodem 3-server, 4-computer, 5-battery, 6electronic compass, 7-GPS receiver, 8-wind sensor Figure 7. Measuring apparatus in fixed position. 1-wind sensors, 2- radiomodem, 3-server, 4-computer omputer, 5-battery 8 przetwornik, 9-doppler scensor, 10 wave probe, 11-container

In figures 8 and 9 photographs examples of the model during drift tests on the lake in condition 3 (ship broken down in the middle) and in condition 4 (stern part drifting separately) are shown, where in figures. 10 and 11 recorded trajectory of the model in those conditions together with its heading and wind force and direction are shown. In the same drawings calculated predicted paths of the model (solid lines) using computer code developed for the sake of comparison are shown (dotted lines). Similar results were obtained for condition 1 and 2 of the model.

Fig.8. Model broken down in the middle during drift tests

Fig. 9. Stern part of the model during drift tests

6. CONCLUSIONS Comparison of the the calculated predicted path of drifting model with experiment where large model in four different damage conditions was drifting freely in natural environmental conditions in the lake reveals reasonable good compliance and it may be concluded that the code may be used for prediction of the drift path of the full scale ship. In this exercise all force coefficients were accurately calculated for the model in damage condition using for hydrodynamic force coefficients and for wave force coefficient results of model tests in the towing tank.

.
Fig.10. Record of trajectory of the drifting model. Model broken-down in the middle.

Fig.11. Record of trajectory of drifting model. Stern part of the model.

Aerodynamic force coefficients were calculated taking into account actual windage area and average wind velocity measured at the model. However current was measured at a distance from the drifting model and was not included in the calculations. When using this methodology to the full scale damaged ship after the casualty all these data will be known only in a very approximate way and the results of prediction may not be equally accurate. It is deemed, however, that the results may be sufficiently accurate for practical purposes and, moreover, the predicted path of drifting ship may be corrected from time to time taking account additional information that may be available from various sources. REFERENCES Anonym (2009): Lekcja bunkierki. Polityka,Nr.49, 05.12 Craneguy Ph. (2000): A simple operational model for the analysis and forecast of pollutant and object drift. Sea-Tech Week Brest 18-20 Dand I.W. , Read A.D. (1978): The drifting vessel problem. Conference on emergency management offshore, EMO 81, Stavanger, Norway 1987. Dand I.W. 1981. Model studies of freely-drifting and towed disabled tankers. Joint RINA/RIN/NI Symposium on disabled large tankers. London Dand I.W.(1981a): The drift of disabled ships in the vicinity of offshore installations: its prediction and correction. National Maritime Institute, Report OT-R8151 Eda H: (1972): Course stability, turning performance and connection force of large barge systems in coastal seaways. Trnsactions SNAME, vol.80, Gusiev A.M. (1954): Wljanje wietra na put i uprawlajemost sudna. Morskoi Transport Hara S. (2000): Towing support tool and objects drift at sea. From Nakhodka to Erika. Exchange of experience in at sea response to offshore oil spills by passing ships. Proceedings of the seminar between Cedre in France and SRI. Brest, Francja. Hara S.(2000a): The Japanese research Project on disabled chips. From Nakhodka to Erika. Exchange of experience in at sea response to offshore oil spills by passing ships. Proceedings of the seminar between Cedre in France and SRI, Brest, Francja Hara S., Hoshino K., Yukawa K., Hasegawa J., Tanizawa K., Ueno M., Yamakawa K.(2000): The OTSS system for drift and response prediction of damaged ships. Japan INA

Hara S., Mitaka S.: The Japanese research project on disabled ships. Presentation at : From Nakhodka to the Erica exchange of experience in at sea response to offshore oil spills by passing ships, Brest, France 6-7 July 2000 Hara S., Yamakawa K., Hoshino K., Yukawa K., Hasegawa J., Tanizawa K., Ueno M. (2004): Development of towing support tool named Optimum Towing Support System (OTSS). Proceedings of the Fourteenth (2004) International Offshore and Polar Engineering Conference. Toulon, France. HAZMAT, (2000): Appendix F. Ship drift analysis. West Coast of North America: Alaska to Southern California. Report 2000-02,. Hoshino K., Hara S., Yamakawa K., Yukawa K. (2004): Estimation method of drift resistance acting on broken or inclined chips. Proc. of the 14th International Offshore and Polar Engineering Conference, Toulon France IMO (1978): Protocol 1978 relating to the SOLAS Convention and to the MARPOL Convention. International Conference on tanker safety and pollution prevention London Inoue S., Hirano M., Kijima K. (!981): Hydrodynamic derivatives on ship manoeuvring. International Shipping Progress Vol.28 No 321. Kobylinski L., Nowicki J., Oleksy J., Jaworski T.: Prediction of drift of disabled tankers Preliminary tests. Proc. XIII International Scientufic and Technical Conference on Marine Traffic Engineering, Malm. Moyano H. (2007): The Prestige case: four years after. TSDT,07 Conference, Newcastle on Tyne. Smeaton G.P. (1981): A mathematical model of the drift of disabled tankers. Joint RINA/RIN/NI Symposium on disabled large tankers ACKNOWNLEDGEMENT The project on prediction of drift path of disabled tankers was executed under the sponsorship of the Ministry of Science and Higher Education, Project Nr. KBN Nr. N N509 1918 33.

Contact details: Lech Kobyliski, lechk@portilawa.com

On the Modelling of Irregular Waves in the Non-linear Ship Dynamics Method Laidyn
Jerzy Matusiak Aalto-University, School of Science and Technology, Marine Technology, Finland

An implementation of irregular waves in the non-linear ship dynamics model Laidyn (Matusiak 2003) is presented. The novelty of the paper lies in the treatment of irregular longcrested waves in the context of the non-linear Froude-Krylov pressures and in the way the linear portion of the sea-keeping response is calculated. The latter is evaluated using the precalculated transfer functions of the response and the actual position of ship in a wave train. Transfer functions are evaluated with a linear strip-theory method. Position of a ship in waves is a result of the solution and thus includes the pertinent non-linearities and time domain solution to the surge, sway and yaw motion.

Keywords: ship dynamics, irregular waves, Froude-Krylov pressures, parametric roll

1. INTRODUCTION Original version of Laidyn (Matusiak, 2000&2001) was meant for evaluation of ship manoeuvring in regular waves. It was based on the so-called two-stage approach. At the first stage of this approach, linear approximation to the rigid body motion in waves is evaluated. A number of non-linearities involved in ship dynamics in waves are taken into account and the total response of ship in the six degrees-of-freedom is solved at the second stage. In particular the non-linearities of the rigid body dynamics, non-linear terms of the restoring and the Froude-Krylov forces, ship resistance, the forces developed by a propulsor and a rudder are taken into account. Details of the method are given for instance in Matusiak (2007). Extension of the method, aimed at dealing with the long-crested irregular waves, led to giving up the concept of the two-stage evaluation of the responses. Instead, a direct solution of ship response is evaluated in the time-domain. All other features of the method are preserved. Similarly as in the original method, some of the forces acting on ship are represented by linear models. In particular radiation forces and moments and also the diffraction part of the wave excitation are given by the linear approximations.

2. SHORT OUTLINE OF THE METHOD For the sake of this paper completeness, a short description of the method is presented in the following. A more detailed description can be found in the abovementioned presentations of the author. Equations of ship rigid motions are given by (Matusiak, 2007)
X g mg sin = m(u + Qw Rv) & Yg + mg cos sin = m(v + Ru Pw) & Z g + mg cos sin = m(v + Pv Qu ) & & & & K g = I x P I xyQ I xz R + ( I z R I zx P I zyQ )Q ( I yQ I yz R I yx P ) R & & & M g = I yx P + I yQ I yz R + ( I x P I xyQ I xz R ) R ( I z R I zx P I zyQ) P & & & N g = I zx P I zyQ + I z R + ( I yQ I yz R I yx P ) P ( I x P I xyQ I xz R )Q where u, v and w are the projections of the velocities of ships centre of gravity in the Earthfixed inertial co-ordinate system on the axes of the moving body-fixed system. The angular position of the ship is given by so-called modified Eulers angles denoted as , and . Refer to Figure 1 for the definitions of the inertia and body-fixed co-ordinate systems. In
(1)

equations 1, Xg, Yg, Zg, Kg, Mg and Ng depict the components of global reaction force and moment vectors acting on the ship. These are given in the-body fixed co-ordinate system xyz. m and Iij mean ships mass and the components of the mass moment of inertia. The relation between the velocities of the ships centre of gravity in the inertial co-ordinate system and their projections u, v and w on the axes of the moving body-fixed system is (Fossen, 1994 and Clayton&Bishop, 1982)
cos cos & X G & YG = sin cos Z & G sin cos sin sin sin cos sin sin sin + cos cos cos sin cos sin cos + sin sin u sin sin cos v cos sin w cos cos
(2)

0 zG Earth fixed axes Gi RG

direction of waves propagation

G G
Horizontal body axes

X,I,U

Y,J,V

,j

,i o ,k o

x,i u

y,j,v

Z,K,W

z,k,w

Figure 1. Co-ordinate systems used to describe ship motion (Matusiak, 2007)

3. IRREGULAR WAVES
Irregular waves met by a ship are generated from their spectral representation in a following way. Given the idealized representation of surface waves in a form of power spectral density

S(), such as for instance the one presented in Figure 2, a time-domain wave train at origin 0 of the Earth fixed co-ordinate system can be formed as

(t) = ai sin i t + bi cos i t = Ai cos ( it i ),

i =1 i =1

(3)

where amplitude of a single wave component having the angular frequency i can be evaluated by integration of wave spectrum as follows

Ai = 2S( i ) . (4) Note that frequency differential can be selected freely. In order to secure a generation of
non-repeatable wave trains from the given wave spectrum, the sample angular frequencies i are random numbers covering the frequency range of interest ( min max). In this case the frequency increments i are also random numbers governed by the adjacent frequencies i and i+1.

Figure 2 Wave spectrum power spectral density of wave amplitude represented by N=100 frequency components.

Wave elevation given by formula (3) can be understood as measured at the origin of the coordinate system, that is X0=0. This wave generation is based on the so-called Deterministic Spectral Amplitude model. In the so-called Non-deterministic Spectral Amplitude model the ai and bi coefficients of the series (3) are taken as a normally distributed uncorrelated variables

(Naito, 1995). This results in the wave amplitude component evaluated with aid of (Matusiak, 2000)

Ai = 2S ( i )
where 22

(5) , 2 is a chi-squared distribution with two degrees of freedom that results from the

2 2

normal distribution of ai and bi coefficients. Both forms of wave generation are available in Laidyn. Jonswap and the so-called two parameter ISSC wave spectra are incorporated in the method.

4. NON-LINEAR HYDROSTATIC AND FROUDE-KRYLOV FORCES AND MOMENTS

When considering ship motions in waves, it is commonly believed that the most important contributors to the non-linearities in the external forces acting on a ship hull are the restoring and Froude-Krylov forces and moments. Evaluation of these is done using a wetted surface of ship hull represented by a discrete panel model. This takes into account both an instantaneous position of ship in space and pressure due to waves extending up to the actual water surface. This is done as follows. Position of each control point c, that is a centre point of a panel, of Figure 3 is transformed from the body-fixed co-ordinate system to the inertial Earth-fixed system using transformation given by Formula 2. Wave elevation above the control point c is given by a sum over the wave components N

(t ) = Ai cos k i (Xc cos Yc sin ) it + i ,

i =1

(6)

where Xc and Yc are the control points of hull surface given in the Earth-fixed co-ordinate system XYZ with an aid of transformation (2) and ki=i/g wave number corresponding to the i-th wave component. For the immersed panel, that is for Zc+(t)>0, pressure is evaluated from the expression
N pc (t ) = g Z c + Ai exp {ki [Z c + (t)]}cos k i (Xc cos Yc sin ) i t + i . i =1

(7)

The argument of exponent function in (7) satisfies the dynamic boundary condition at the actual water surface. The pressure profile thus obtained is called sometimes as the stretched one. This approach is a kind of an extension of the linear wave theory to incorporate non-linear effects associated with the variation of ships wetted surface. Figure 3 below illustrates the quantities used in evaluating hydrostatic and Froude-Krylov pressures.

Fig. 3 Evaluation of hydrostatic and Froude-Krylov pressure.

The forces F and moments M are obtained by integrating the pressure (8) in the body fixed coordinate system. This integration is performed numerically by summing up the contribution from each wetted panel using
total total FF.K = FF.K;i = pi Si n i i i total F.K;i M M

total F.K

= ri F
i

, ,

(9)

where the total number of the panels is denoted by M, Si is panel area, ni unit vector normal to panel and ri the position vector of the control point in the body fixed co-ordinate system xyz.

5. LINEAR APPROXIMATION TO SHIP MOTIONS Linear approximation to the global responses of ship in irregular waves is evaluated in order to judge the effects of non-linearity on the derived responses. Normally, in the linear

seakeeping theory, a constant forward speed is assumed. In the Laidyn method surge motion of ship is evaluated in the time domain taking into account amongst the others propeller action and variations of the wetted surface. Thus in-plane motion of ship is simulated in timedomain along with the other motion components. This results in ship position XG,YG in the Earth-fixed co-ordinate system. This and the knowledge of transfer functions of the corresponding responses make it possible to evaluate linear approximation of the responses using the expressions

xL (t ) = Ai x L 0 ( i )cos k i (XG cos YG sin ) i t + i x,i

i =1 N

yL (t ) = Ai yL 0 ( i )cos k i (XG cos YG sin ) i t + i y,i

i =1 N

zL (t ) = Ai zL 0 ( i )cos k i (XG cos YG sin ) i t + i z,i

i =1 N

L (t ) = k i Ai L 0 ( i )cos k i (XG cos YG sin ) i t + i ,i

i =1 N

(10)

L (t ) = k i Ai L 0 ( i )cos k i (XG cos YG sin ) i t + i ,i

i =1

L (t ) = k i Ai L 0 ( i )cos k i (XG cos YG sin ) i t + i ,i ,

where terms with subscripts L0 depict gain factors of the transfer functions and corresponding phase angles. 6. EXAMPLE RESPONSES OF SHIP A1 The newest benchmark study initiated by the International Towing Tank Conference and presented in (SAFEDOOR, 2008) concerned the so-called ship A-1. This is a containership of waterline length of LPP = 150 m. A model of this ship was also investigated earlier in a similar study (ITTC, 2002). A number of model tests, mainly in head and in the following regular waves, were conducted. One of the simulated tests is presented in the following. This test (depicted T16) was conducted in the following regular waves with a ship running at Fn=0.04 and wave amplitude A=6 m and period T=8 s. Simulated result in a form of roll motion is given in Figure 4 below. A clear development of parametric roll resonance is visible. Amplitude of this motion being approximately 150 is lower than the measured one (260). It is worth noting that in waves of
i =1

amplitude 3.6 m, both in model tests and in the simulations, roll motion did not occur for the same ship speed and wave length (test T15).
Test T15 20 15 10 5 0 480 -5 -10 -15 -20 Time [s]
Wave [m] Roll [deg]

540

600

660

720

780

Fig. 4 Ship A1 rolls in the following waves of amplitude A=6 m and period T=8 s. Ship speed is Fn=0.04 and megacentric height GM0=1 m.

Test T19 was conducted in the following irregular waves given by the Jonswap spectrum of significant wave height Hs= 5 m and peak period Tpeak=8 s. Ship speed was Fn=0.08. Ship model did not roll in this test, neither it rolled in the simulations. In order to initialize ship rolling, simulations were done for stern quartering waves (heading =300) and speed Fn=0.235. Result of the simulated roll motion is given in Figure 5. Simulated case is critical for the ship in this particular case because of the focusing effect of waves. The encounter period of majority of waves is very close to the natural period of roll (T=21 s). As a result a resonant roll motion develops. Time domain solution of roll yields substantially higher roll amplitudes than the linear approximation (depicted in Figure 5 by roll_l). In Figure 6 pitch motion of the same time period is shown. Both, the directly solved pitch and its linear approximation (pitch_l) are presented. In this case a good agreement of both solutions is noted.

stern quartering irregular seas 10 8 6 4 2 0 -2780 -4 -6 -8 -10 -12 Time [s] ]Wave [m ]Roll [deg roll_l 840 900 960 1020 1080 1140 1200

Fig. 5 Ship A1 rolls in the stern following irregular waves (Jonswap spectrum of Hs=5 m and Tpeak=8 s). Ship speed is Fn=0.235 and megacentric height GM0=1 m.
stern quartering irregular seas 1.5 1 0.5 0 780 -0.5 -1 -1.5 Time [s] ]Pitch [deg pitch_l

840

900

960

1020

1080

1140

1200

Fig. 6 Pitch motion of ship A1 rolls in the stern following irregular waves (Jonswap spectrum of Hs=5 m and Tpeak=8 s). Ship speed is Fn=0.235 and megacentric height GM0=1 m.

7. CONCLUSIONS Inclusion of the irregular waves modelling in the Laidyn method preserved main features of the method. In particular capability to predict parametric roll resonance is preserved. Evaluation of resonant roll response in stern quartering seas at high speeds requires further studies. In particular validating model tests would be a valuable tool to check reliability of the predictions. REFERENCES Clayton B.R. and Bishop R.E.D (1982) Mechanics Of Marine Vehicles, London, ISBN 0 419 12110-2. Fossen, T.,I. (1994) Guidance And Control Of Ocean Vehicles, J. Wiley&Sons, ISBN 0 471 94113 1. Matusiak, J. (2000) Two-Stage Approach To Determination Of Non-Linear Motions Of Ship In Waves, 4th Osaka Colloquium on Seakeeping Performance of Ships, Osaka, Japan, 17-21st October, 2000. Matusiak, J. (2000a) Dynamics of cargo shift onboard a ship in irregular beam waves, International Shipbuilding Progress, 47, No 449 (2000) pp. 77-93 Matusiak, J. (2001) Importance Of Memory Effect For Capsizing Prediction, 5th International Workshop On Ship Stability, University of Trieste 12. - 13. September 2001 ITTC (2202) 23rd International Towing Tank Conference, Report of the Specialist Committee of Extreme Ship Motions and Capsizing, Venice Italy, September 8-14, 2002, in Proceedings of the Conference, Vol. 2. Matusiak, Jerzy; (2003) On the effects of wave amplitude, damping and initial conditions on the parametric roll resonance. Proceedings of the 8th International Conference on Stability of Ships and Ocean Vehicles, Madrid, Spain, September 2003, pp. 341-348 Matusiak, Jerzy; (2007) On certain types of ship response disclosed by the two-stage approach to ship dynamics. ACME, Archives of Civil and Mechanical Engineering, 2007. Vol. VII, nro 4, 151-166 Naito, S. (1995), Generation and absorption of waves. Symposium on Wave Generation, Analysis and Related Problems in Experimental Tanks- Especially on Directional Wave. Yokohama National University, 25 September 1995. Pp. 1-27.

SAFEDOR

(2008)

Benchmark

Study

on

Parametric

Rolling;

in

http://www.naval.ntua.gr/~sdl/sibs/intact.htm CONTACT DETAILS Jerzy Matusiak; Aalto University, School of Science and Technology Jerzy.Matusiak@TKK.Fi

Probability of Capsizing in Dead Ship Condition

Tomasz Hinz Polish Register of Shipping, Gdask, Poland

The nature of the contemporary intact stability regulations is prescriptive. They are based mostly on requirements to the righting arm curve (GZ curve) on calm water. This approach is further development of Rahola's work which was conducted in the 1930s. The first intact stability resolutions were approved by IMO in the 1960s. At present, IMO works on development of new regulations which are to be largely based on the probability approach. The SLF subcommittee proposed four main capsizing scenarios: dead ship condition, pure loss of stability on wave crest, parametric roll and surf-riding with broaching. These scenarios are the starting points for changing the current regulations. Many scientific centres have undertaken the work on creating the new stability criteria. This paper presents the computations of the capsizing probability for RoPax ship in the case of Dead Ship Condition scenario. The non-linear model of ship motion for regular wave was applied in the calculations.

Keywords: capsizing, dead ship condition, probability

1. NEW INTACT STABILITY CRITERIA BASED ON THE PROBABILITY METHODS In 1968 the intact stability rules were approved as the A.167 resolution (ES. IV). Then, in 1993, this resolution was replaced by another one A.749 (18). In July 2010, the new revision of IS Code is to come into force. During the 85th session of Martine Safety Committee (MSC) it was agreed that part A of IS Code will be obligatory as it is defined in SOLAS Convention. However, all above mentioned amendments are not followed by any crucial changes in the text of IS Code. The Code is still based on the same assumptions, according to which the ship indicator of stability safety is the righting arms curve on calm water. The discussion about new generation of stability rules took place during the 50th (SLF 50/4, 2006), 51st (SLF 51/4, 2008) and 52nd (SLF 52/3, 2009) sessions of the Sub-committee on Stability, Load Lines and on Fishing Vessels (SLF) of IMO. It was proposed that the new generation of regulations should be based on four intact stability failure scenarios. Dead Ship Condition, i.e. ship without forward speed, exposed to action of waves and wind Pure-loss of stability Parametric roll Surf-riding and broaching

In order to present a new approach to the stability issue, it is necessary to define the intact stability failure. The definition can be found in SLF document 51/WP.2 (SLF 52/WP.2, 2008): Intact stability failure is a state of inability of a ship to remain within design limits of roll angle and combination of rigid body accelerations. Two types of the intact stability failure can be recognized (SLF 50/4/4, 2006): Partial intact stability failure the partial loss of the operation abilities of a ship, additionally combined with the potential danger for people as well as for cargo and equipment. Total intact stability failure the total loss of a ship, additionally combined with the loss of lives.

A process of determining the new intact stability regulation began during the 50th session of SLF IMO subcommittee. The delegations from Japan, the Netherlands, the USA (SLF 50/4/4, 2006) and from Italy (SLF 50/4/12, 2006) presented their conceptions concerning the new intact stability regulations. The subcommittee presented four definitions of criteria which deal with the assessment of the intact stability failure in different ways: probabilistic performance-based this criterion is based on the physical model of intact stability failure considering the probability of an event; deterministic performance-based this criterion resembles the previous one, except the fact that an event does not have the probability nature but is the determined one; probabilistic parametric criterion this criterion is based on the value measurement connected with an occurrence but it does not encompass the physical model of the occurrence. It exploits one or more stochastic values; deterministic parametric criterion this criterion also does not contain the physical model but is based on one or more deterministic values which take part in measuring the values connected with the occurrence. This criterion is applied to present regulations. The new generation of stability regulation is to be a multi-stage process connected with two types of criteria (additionally IS Code is used): Performance based criteria criteria combined with attaining the performance; they can be based on the model test or numerical simulations and can have probabilistic or deterministic nature; Vulnerability criteria criteria connected with a ship susceptibility to intact stability failure for particular scenario; these criteria can be divided into two or more levels: level 1 is based on the simple criterion combined with the ship geometry level 2 is based on the simple physical model of an occurrence of dangerous phenomenon Figure 1 presents the diagram of the concept of new stability regulations. Two levels of criteria, vulnerability level and the one based on performance, are different for disparate ship capsizing scenarios.

Figure 1. Process of assessing in the new generation of stability (SLF 52/3/3, 2009)

The probability of ship capsizing is the main part of the probabilistic criteria. The model which is described by the formula below is the most popular probability model (Mc Taggart, 1995) (Mc Taggart, 1998), (Mc Taggart, 2000), (Lee et al., 2009):

(1)

where: probability of ship capsizing during one year ship speed heading wave steepness maximal significant height of wave

minimal significant height of wave which causes the capsizing in the presence of parametrical data

The formula presented is valid but only and exclusively under of the assumption that the heading and ship speed are independent of wave condition. This assumption is not totally correct in the light of existing recommendations for captains to change the course and the speed depending on the weather conditions. The probability is often calculated by the Monte Carlo method (Mc Taggart, 2000). However, not only the probability of ship capsizing can be calculated this way, but also the probability of the critical maximum roll angle .

Instead of calculating the probability of capsizing in the direct way, Mc Taggart (2000) suggested using the method based on the probability of the critical roll angle. This method is based on the calculation of conditions. According to this approach, it is possible to calculate the probability of capsizing by applying the formula (2): the distribution of the critical roll angel for different sea

(2)

2. THE PROBABILITY OF SHIP CAPSIZING IN DEAD SHIP CONDITION Monte Carlo method was used to calculate the probability of ship capsizing for the Dead Ship Condition scenario. It was conducted with reference to the probability of critical maximum roll angle, where, the capsizing probability was defined with the usage of formula 2. Monte Carlo method can be described with the formula 3: (3)

number of simulations where the intended angle was reached number of all simulations

LAIDYN program was used to conduct all simulations. The program was created and developed at Aalto University in Espoo, Finland (Matusiak, 2000), (Matusiak, 2002). LAIDYN method is based on the assumption that the complete answer of a ship equals the sum of linear and non-linear parts. Such division results from the fact that the linear computation methods are well-known. It causes the situation where the radiation and diffractive forces are presented by linear equations quite well. In this method, the main part of the first order load is calculated with the linear approximation, based on the current heading and location in relation to a wave. Defining the non-linear part, such elements as non-linearity as a result of ship shape, hydrostatics, wave force, propeller and ruder forces, were taken into consideration.

3. COMPUTATIONS AND RESULTS RoPax type of ship was chosen as the example for the calculations of capsizing probability for the Dead Ship Condition scenario. The basic parameters for this type of ship are presented in Figure 2. The centre of gravity of the ship was located on the height ZG = 11.5 [m].

Figure 2. Lines drawing for RoPax ship (Mattila, 1999)

The sea wave statistics data were chosen for the seas around the European continent and were taken from Global Wave Statistics BMT Ltd (1986). All, together 66911 computations of ship motions in waves were conducted. This sum matches about 1/3 life of a ship. The analysis of results shows that the increase of computation number for particular ship does not affect the scale of probability significantly.

Table 1. Probability of capsizing for Dead Ship Condition scenario

Critical roll Angel 25 30 35 45 60 75

Probability of reaching the critical angel 1.971E-3 1.421E-3 5.886E-4 3.286E-4 1.196E-4 1.928E-5

Table 1 presents the results of the probability computations for different angles where the LOSA accident can appear. The angel of hell equal 60 deg has been chosen as the critical heel in the final estimation of capsizing probability.

4. CONCLUSIONS The probability of capsizing of RoPax ship during one year operation is found to be 1.196E-4. This is very low probability which means that the capsizing may occur once in 10000 years for single vessel, or one ship in populations 10000 ships during in one year period. This is very low probability which may be considered as acceptable. The Dead Ship Condition scenario is considered at IMO as one of main scenarios for stability accidents. It is assumed as especially dangerous for ships with big superstructures, e.g. RoPax or loaded container ship. The results of the computations performed here do not support this opinion. Monte Carlo method as a tool for the probability computations has some disadvantages. The most serious one is its time-consuming nature. This method gives better results when the number of computations increases.

ACKNOWLEDGEMENTS Special acknowledgements for prof. Matusiak from Aalto University for providing program LAIDYN.

REFERENCES Bassler, C. C., Belenky, V., Bulian, G., Francescutto, A., Spyrout, K., Umeda,N., 2009, A review of available methods for application to secend level vulnerability criteria. In: 10th International Conference on Stability of Ships and Ocean Vehicles. pp. 111_128. BMT Ltd (Ed.), 1986. Global Wave Statistics, BMT Ltd, London. Lee, S.-K., Long, Z.-J., Jeong, J.-H., Cheon, S.-J., 2009. Risk assessment method of simulation-based for the intact ship stability, In: Proceedings of ICCES 09. Mattila, M., 1999. Experimental study on transverse stability of fast RoPax vessels in waves. Master's thesis, Helsinki University of Technology, Espoo, Finland. Matusiak, J., 2000. Two-stage approach to determination of non-linear motions of ship in waves. In: 4th Osaka Colloqium on Seakeeping Performance of Ships. Osaka University, Osaka Prefecture University, pp. 326_332. Matusiak, J., 2002. Towards an unified theoretical model of ship dynamics, In: The Maritime Research Seminar. Sjokulla, Finland. Matusiak, J., 2007. On certain types of ship responses disclosed by the two stage approach to ship dynamics Archives of Civil and Mechanical Engineering VII (4). McTaggart, K., 1995. Wind effects on ship capsize risk, Tech. rep., National Defence. Research and Development Branch. McTaggart, K., 1998. Ongoing work examination capsizing risk of intact frigates using time domain simulation, In: Proceedings of 4th International Stability Workshop. McTaggart, K., 2000. Improved modelling of capsize risk in random seas, Tech. report, National Defence. Research and Development Branch.

SLF 50/4, 2006. Revision of the intact stability code. Report of the working group on intact stability at SLF49 (part 2). Submitted by the chairman of the working group, In: 50st session of IMO Sub-Committee on Stability and Load Lines and on Fishing Vessels Safety. IMO SLF 50/4/12, 2006. Comments on SLF 50/4/4. Submitted by Italy. In: 50st session of IMO Sub-Committee on Stability and Load Lines and on Fishing Vessels Safety. IMO SLF 50/4/4, 2006. Framework for the development of new generation criteria for intact stability. Submitted by Japan, Netherlands and USA. In: 50st session of IMO SubCommittee on Stability and Load Lines and on Fishing Vessels Safety. IMO SLF 51/4, 2008. Revision of the intact stability code. report of the working group (part 2). Submitted by the chairman of the working group. In: 51st session of IMO Sub-Committee on Stability and Load Lines and on Fishing Vessels Safety. IMO, p. 5. SLF 52/3, 2009. Revision of the intact stability code. report of the working group (part 2). Submitted by the chairman of the working group. In: 52st session of IMO Sub-Committee on Stability and Load Lines and on Fishing Vessels Safety. IMO SLF 52/3/3, 2009. Two-layered vulnerability criteria and direct assessment of ships stability Submitted by Poland. In: 52st session of IMO Sub-Committee on Stability and Load Lines and on Fishing Vessels Safety. IMO SLF 52/WP.2, 2003. Revision of the intact stability code. In: 51st session of IMO SubCommittee on Stability and Load Lines and on Fishing Vessels Safety. IMO

CONTACT DETAILS Tomasz Hinz, Polish Register of Shipping, Gdask, Poland tomasz.hinz@prs.pl