You are on page 1of 76

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


The Propulsion Committee

Final Report and Recommendations to the 23rd ITTC

1. GENERAL 1.1. Membership and Meetings The members of the Propulsion Committee of the 23rd ITTC were as follows: Neil Bose Christian Dugu (active until June 2001) Pier Giorgio Esposito (to April 2001) Jan Holtrop Stuart Jessup Jin-Tae Lee Friedrich Mewis Alexander Poustoshny Francesco Salvatore (from May 2001) Yasushi Shirose Stuart Jessup was elected Chairman by the conference. Friedrich Mewis was elected Secretary by the Committee. The Meetings of the Committee were held as follows: Rome, Italy St. Johns, Canada St. Petersburg, Russia Hamburg, Germany Feb. 2000(8) Oct. 2000 June 2001 Jan. 2002 (8) (8) (5)

1.2. Recommendations of the 22nd ITTC The recommendations for the work of the Propulsion Committee as given by the 22nd ITTC were as follows: 1. Review the state-of-the-art, comment on the potential impact of new developments on the ITTC, identify the need for research and development in the areas of propulsors, cavitation and powering performance. Monitor and follow the development of new experimental techniques and extrapolation methods. 2. Review the ITTC recommended procedures, benchmark data, and test cases for validation and uncertainty analyses and update as required. In particular, the following procedures should be reviewed: Model Scale Cavitation Pattern Tests ITTC Procedure 4.9-0.3-03-03.1 Description of Cavitation Appearances ITTC Procedure 4.9-03-03-03.2 3. Develop a procedure for predicting the performance of ships with azimuthing thrusters as the main propulsor. 4. Identify the requirements for new procedures, benchmark data, validation, uncertainty analyses and stimulate the necessary research for their preparation.


The Propulsion Committee

23rd International Towing Tank Conference

5. Review methods for scale effects on the passive components of propulsors and for assessing screw propeller scale effects with emphasis on the occurrence of excessive laminar flow. 6. Review the development of numerical design and analysis methods for propulsors. Follow the developments in the modelling of unconventional and multi-component propulsors. 7. Review developments in experimental techniques and analytic methods for modelling the propulsive effects of propellerrudder interaction including cavitation and cavitation effects. 8. Review developments in analytic and experimental methods for hydroelastic phenomena and propulsors and recommend procedures to account for hydroelastic effects in predicting and evaluating propulsor performance. 9. Prepare an up-to-date bibliography of relevant technical papers and reports. 1.3. General Remarks The Propulsion Committees work included preparation and revision of ITTC Quality Manual procedures, the development of a potential new procedure, and the execution of the traditional ITTC subject review tasks. A major part of the Committees efforts related to advancement and model testing of podded propulsors. The revision of the QM cavitation procedures, initially appeared straightforward, but ultimately involved considerable effort and coordination with the related cavitation specialist Committees and the QS group. The cavitation procedure, 4.9-03-03-03.1, Model Scale Cavitation Pattern Test was revised slightly, and was renamed as the Model-Scale Cavitation Test, new procedure No. 7.5-02-03-03.1. The

new title reflected a broader, basic description of the general cavitation test. Areas covered by the Specialist Committees related to cavitation induced hull pressure and water quality were not included and were expected to be addressed in new procedures. Procedure 4.9-03-03-03.2, Description of Cavitation Appearances, was updated with new photographs and description of cavitation types, new procedure No. 7.5-0203-03.2. The creation of the procedure, Podded Propulsion Tests and Extrapolation Methods, involved considerable effort. The procedure is considered a work in progress. With the rapid advancement of podded propulsion, the development of test and prediction methods is presently evolving. The Committee believes that the procedure could be improved and revised by later Committees. In section 5, under task 4, the Committee has continued to support the use of a physically based friction line as an eventual replacement of the ITTC 57 correlation line. Recent analysis has supported use of the Grigson friction line. The Committee originally planned its fourth meeting in October 2001 in Hamburg, Germany. Because of the September 11th terrorists attacks, the meeting was postponed to January 2002. To compensate for the delay, the Committee held a virtual meeting. Members emailed their task contributions by a specified date, and this was followed with a weeklong comment period, where drafts were revised based on other members input. The end result was a prepared full report draft, which was reviewed at our January meeting. The virtual meeting was not a replacement for a regular, attended meeting, but it did successfully accelerate progress in completing the Committees report.

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


2. REVIEW THE STATE-OF-THE-ART, COMMENT ON THE POTENTIAL IMPACT OF NEW DEVELOPMENTS ON THE ITTC, IDENTIFY THE NEED FOR RESEARCH AND DEVELOPMENT IN THE AREAS OF PROPULSORS, CAVITATION AND POWERING PERFORMANCE. MONITOR AND FOLLOW THE DEVELOPMENT OF NEW EXPERIMENTAL TECHNIQUES AND EXTRAPOLATION METHODS 2.1. Surface Piercing Propellers During the last decades the interest of shipbuilders in the application of surface piercing propellers (SPP) to high-speed craft grew significantly worldwide. At present, SPPs are recognized as an outstanding propulsion device for small planing or semi-displacement craft. SPPs allow the attainment of high efficiency at high ship speed at relatively high rpm. Ventilation due to the cyclic blade entry into the water almost completely prevents the occurrence of vapour cavitation. In addition, shaft, struts and propeller hub emersion yields a strong reduction in appendage drag. A thorough investigation of SPP flow must include freesurface analysis, aeration, cavitation, viscosity, surface tension effects and transient phenomena during blade entry into the water and exit from it. Reviews of published experimental data show that in spite of continued investigations carried out for many years there is still no universal approach for establishing principal SPP non-dimensional parameters. The 22nd ITTC Specialist Committee on Unconventional Propulsors identified two parameters relevant for extrapolating SPP model test results to fullscale conditions: Froude and Weber numbers. In the literature on SPPs, the Froude number is defined using several different character-

istic lengths. Alternatives include: propeller diameter (Basin, 1958), (Basin & Goshev, 1963), (Bolotin, 1991), (Egorov, 1932), (Iljin, 1975), propeller shaft depth (Gutsche, 1967), lower blade edge depth (Ferrando, 1997), centre of gravity depth (Korshunov & Eroshevsky, 1984) and the square root of the submerged propeller disk area. Non-dimensional hydrodynamic load coefficients are calculated both as in the case of totally submerged propellers (Nagau, 1974) and by taking into account the area of the submerged portion of the propeller disk (Ferrando et al., 2001). It is apparent that there is a strong need to formulate standard definitions of SPP non-dimensional parameters in order to make possible correct comparisons of data from SPP model tests, and to facilitate practical design applications. The 22nd ITTC Specialist Committee on Unconventional Propulsors highlighted two major topics relevant for further research on SPP: extrapolation methods of open-water tests; extrapolation methods of propeller-hull interaction.

In the case of open-water measurements, it is generally agreed that the standard Reynolds criterion is generally applicable. The determination of threshold values above which Froude number and Weber number scaling effects of open water testing are negligible is important (Ferrando, 1997). Moreover, there is some evidence of the importance of the cavitation number, especially for propulsion predictions. Some researchers claim that under developed ventilation conditions the influence of the cavitation number upon the SPP hydrodynamic performance is insignificant (Brandt, 1973; Rose & Kruppa, 1991). However, other results indicate significant relationships between these parameters, as shown in Figure 2.1 (Prischemihin & Korshunov, 1969).


The Propulsion Committee

23rd International Towing Tank Conference

Full scale extrapolation of propeller-hull interaction data is made difficult by the almost total absence of sea trials results. This is essentially due to budget limitations that are characteristic of SPP-propelled craft design. Model test experience reveals that sometimes this interaction can be as significant as to necessitate correcting propulsor geometry after sea trials of the built ship. The 22nd ITTC Specialist Committee on Unconventional Propulsors briefly considered the problem and found the SPP thrust deduction coefficient to be negligible or zero. The conclusion of this Committee was: Considering that surface piercing propulsion does not include any additional wake modifying device and that the physics of the wake production process is the same as in the case of conventional propulsion, there are no obstacles to using the same scaling procedure as that used in the case of conventional propeller arrangements. It has also been observed that in some cases SPPs induce a change in the dynamic trim of the ship, which impacts resistance and propeller operation as affected by the SPP immersion depth. The above effects should be considered when predicting propulsion performance.
10KQ 1,0 0,8 0,6 0,4
Fn=6.2; Kn=5.6 Fn=12.5; Kn=14 Fn=6.2; Kn=3.6 Fn=6.2; Kn=5.6

Recent developments on scaling performance data from open-water tests is addressed by Ferrando et al. (2001). A family of four parent four-bladed propellers with pitch ratios of 0.8, 1.0, 1.2, 1.4 was considered. Model tests were performed in the cavitation tunnel of the University of Genoa, in the towing tank of the University of Naples, and in the towing tank of IMD in St. Johns, Newfoundland. A modified advance ratio J is introduced in order to normalize data obtained by using different values of the shaft angle . In addition, a modified Weber number is introduced to obtain independency of values from immersion depth. Respectively,
J = V A

n 2 D 3 IT cos , Wn ' = nD K

The symbol IT denotes the immersion coefficient, equal to the ratio between the maximum blade tip immersion hT, measured from the undisturbed water surface in nominal conditions, and the propeller diameter D. The symbol K is the kinematic capillarity defined as the ratio: surface tension/density. Non-linear regression curves are determined to fit measured values of J corresponding to complete ventilation of the propeller blade as a function of Wn' and of the pitch ratio P/D. The result is shown in Figure 2.2, where JCR denotes J values corresponding to transition from partial to complete ventilation.

Fn=12.5; Kn=0.92 Fn=6.2; Kn=3.6 Fn=12.5; Kn=0.92 Fn=12.5; Kn=14


0,0 0,0





1,0 J 1,2

Figure 2.1 SPP hydrodynamic characteristics nD versus Froude number Fn = and cavitaghS



p pV (hS=0.5D depth 0.5h 2 D 2 of submersion of a propeller shaft axis).

tion number K n =

Figure 2.2 Regression curves for JCR of SPP models with various pitch ratios (Ferrando et al., 2001).

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


Both Wn' and P/D influence JCR. In particular, the influence of Wn' on JCR decreases with increasing Wn' and a threshold value upon which JCR does not depend on Wn' may be found. Such a threshold is roughly linearly dependent on P/D. Plotting curves in the (JCR Wn') plane may be helpful to determine the minimum advance coefficient above which model tests represent full scale behaviour. The influence of the immersion ratio IT is also addressed. The analysis is limited to advance values above the transition, where the waterline is stable and the immersion depth is clearly detected. Below the transitional coefficient the waterline is highly unstable and it is not possible to uniquely quantify the immersed area. Modified thrust and torque coefficients are proposed where the nominal immersed propeller disk area A0 referred to hT is used as the reference area:
K 'T = Q T , K 'Q = 2 2 n D A0 n D 3 A0

be based only in part on approaches that have been developed for submerged propeller flow. In particular, the assessment of methods for practical design is far from being achieved.


Figure 2.4 Regression curves for K'Q versus J of SPP models with various pitch ratios (Ferrando et al., 2001). A numerical analysis of SPPs flow is proposed by Young & Kinnas (2000). A hydrodynamic analysis SPP code has been developed by modifying PROPCAV, a boundary element code developed for the analysis of unsteady partial sheet cavitation on fully submerged propellers. A general model of the SPP hydrodynamic problem may be seen in Figure 2.5. Boundary conditions on the free water surface and on the ventilated cavity surface are imposed by assuming that pressure is constant and equal to the atmospheric pressure. A linearized boundary condition at the free surface based on an infinite Froude number assumption is used. No attempt is made to model free-surface elevation variations. The methodology is not able to take into account finite-thickness blade trailing edge geometry, which is usual in current SPP design. Calculations by PROPCAV are performed by using a modified blade geometry with zero-thickness trailing edge. Predicted maximum blade loading shows a reasonable agreement with experiments, but large discrepancies exist at both blade entry into the water and exit from it (see Figure 2.6).

If J is greater than JCR, both K'T and K'Q are almost independent of the immersion coefficient. Non-linear regression curves are determined to represent such force coefficients as functions of J and P/D. Results are given in Figures 2.3 and 2.4.


Figure 2.3 Regression curves for K'T versus J of SPP models with various pitch ratios (Ferrando et al., 2001). Due to the complex nature of SPP flows, theoretical modelling is complicated and may


The Propulsion Committee

23rd International Towing Tank Conference

An application of the surface piercing propeller concept was addressed by Ferrando et al. (2000). This paper describes the application of SPPs to low-speed merchant ships. For this particular case, the terminology Partially Submerged Propeller (PSP) is preferred rather than SPP. The basic idea underlying a PSP is that if partial emersion of the propeller blades is allowed, a higher propeller diameter may be used and hence an increased hydrodynamic efficiency should be achieved. Experimental investigations have been conducted to analyze the influence of advance coefficient, immersion coefficient, Weber, Reynolds and Froude numbers, and of the yaw angle. Resistance, self-propulsion tests and wake measurements have been performed at Naples University Model Basin on a model of a small tanker in ballast and full load conditions. Measurements performed by using a conventional propeller (CP) with D = 3 m, and a PSP with D = 5.07 m (at full scale) are compared. Due to the larger diameter, the housing of the PSP required a modification of the stern lines with respect to the model equipped with the conventional propeller. The modified geometry determined an increase of model resistance with respect to the original design. Values of the immersion ratio used during tests and some measured quantities are summarized in Table 2.1, where the symbol denotes the difference in percent between CP and PSP data. The use of PSP allows a marked increase of the propulsive efficiency and a reduction of the power in both full load and ballast conditions. Ferrando et al. (2000) also report the experience of a 5500 tons livestock carrier design with PSP installed. Sea trials confirmed performance improvements predicted by model test measurements. During the design process, a computer program based on the lifting-line approach was used. The model takes into account the reduction of water density at certain blade positions due to the presence of air. Numerical results demonstrate a

Figure 2.5 Definition of exact and approximated flow boundary.

Figure 2.6 SPP blade force predictions compared to experiments. Propeller model 841-B (Young & Kinnas, 2000). The authors conclude that the following aspects require further study: blade vibration due to the cyclic loading of the blade associated with free water -surface crossings; changes of free-surface elevation due to blade entry and exit; modelling of separated flow behind thick trailing edges.

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


satisfactory account of the effects of cyclic blade emersion on propeller performance. Table 2.1 Comparison between totally and partially submerged propellers (Ferrando et al., 2000).
Full load CP IT 1.86 RT [t] 15.02 0.545 0 1-w 0.711 BHP 2240 PSP 1.13 16.61 0.683 0.721 2056 Ballast

In addition, it is necessary to develop a method for extrapolation of model test results to full-scale conditions. A possible option is represented by the model test procedure for SPP design and ship propulsion prediction currently in use at KSRI. It includes:


CP PSP (%) 1.38 0.84 +10.6 11.41 13.69 +20.0 +25.3 0.552 0.696 +26.1 +1.4 0.657 0.715 +8.8 -8.2 1664 1611 -3.2

cavitation tank open-water SPP model tests at different propeller shaft axis depths to obtain SPP hydrodynamic performance data comparable with other screw propellers; towing tank self-propelled model tests to determine effective thrust-power curves versus propeller revolutions and ship speeds without cavitation; cavitation tank SPP + dummy hull model tests in oblique flow to define the influence of the cavitation number and of the free water surface configuration; blade strength analysis.

The status of SPP research outlined above points out some priorities. Standardization of hydrodynamic non-dimensional parameters is necessary to correctly compare various SPP model test data and their application in practical design. Existing research experience allows the development of model test methods which could give results that are in close agreement to full-scale SPP characteristics. Considering the current interest in such propulsors, these tasks appear to be rather urgent. In order to achieve this goal it will be necessary to formulate and resolve a number of problems:

2.2. Propulsion of Single Screw MegaContainer ships

establish a list of geometrical, kinematic and hydrodynamic similarity criteria that are undoubtedly necessary for both SPP open-water and self-propelled model tests; determine the range of hydrodynamic similarity criterion values for which it is possible to provide suitable correspondence of model results to full-scale conditions; select a rational unified format of dimensionless numbers and coefficients that would be most convenient for SPP design practices and best representative of physical conditions of SPP operation; establish the necessary scope of data to be obtained from model tests.

The last few years have seen the advent of very large single-screw container ships of extremely high power, running at speeds of over 25 knots (e.g. Anon, 2000). The capacity of these ships exceeds 6000 TEU, whereas trends point towards even larger ships with propulsive powers approaching 100 MW on a single shaft (Pederson, 1999). Propellers have 6 blades, a large blade-area ratio (>0.9) and diameters in the range >8.75 m. All of these propellers are driven by slow-speed, twostroke, crosshead Diesel engines, operating at a rotation rate of approximately 100 rpm. The high specific loading of these propellers (>1 MW/m2 disc area; CT 1.0) and the high circumferential tip speeds (>45 m/s) indicate that the design margins of these propellers are extremely small and that successful propellers can


The Propulsion Committee

23rd International Towing Tank Conference

only be designed for ships hulls which generate a wake distribution that is sufficiently homogeneous. As a consequence, the development of propellers for this class of ships is extremely critical even though the thrust loading coefficient has a normal value. So, hydrodynamic prediction tools, either experimental or computational, are required to be not only able to distinguish between alternatives but able to accurately predict the absolute level of the vibration excitation and the risk of cavitation erosion. From several systematic series of experiments at MARIN on single-screw hull forms typical of container ships the following Difficulty Index DI has been devised. At the preliminary design stage, the main parameters including the tip-hull clearance, the allowable depth of the central wake peak and some of the main propeller dimensions may be assessed:
DI = T N 2 w5 3/4 5 10

stages of the design is essential to arrive at successful propulsion solutions for these highpowered, large single-screw container ships. Propeller modifications appear to have only a secondary effect on the cavitation performance once an unfavourable wake pattern is established. Of special concern is the rotation rate offered by the engine manufacturers. At the largest powers available, rotative speeds of the engine are about 100 rpm. This results in circumferential tip speeds on the order of 46-48 m/s for the maximum power currently available (about 68 MW). The single-screw wake distributions and the advent of even larger powers results in an increase in the number of cylinders, requires even larger propellers, and indicates that somewhat lower rotative speeds could probably be more attractive. Regarding the design of the propellers, special attention is to be given to the radial load distribution, the skew and the rake. There is no universal opinion and experience on the use of special sectional profiles. A quite interesting procedure is that of Kawakita & Hoshino (1998) who claim superior cavitation properties and lower hull pressure fluctuations to be attained by the SUMT optimization technique applied to various propellers, including one of a container ship. On extremely high-powered container ships the extent and cavity volume variations are quite large. If the cavitation would have been of a more limited nature the substitution of standard sectional profiles by those optimised from the viewpoint of cavitation inception would certainly lead to a significant reduction of the cavitation extent and the amplitudes of the hull-pressure fluctuations. There are indications, however, that on heavily cavitating propellers, such as those for the high-powered container ships, the use of sectional profiles optimised for the inception of cavitation may lead to sections with a higher risk of erosion and higher fluctuations in the cavitation once the cavitation is fully developed. Apparently, these sections are not universally successful. At present, the advantage of

Z Ae / A0 C


the propeller tip radius and it is defined as:

w expresses the depth of the wake peak at

w = (V max V min )/ V


w = w max w min
C = tip-hull clearance in m N = propeller rotation rate in rpm Z = number of blades T = propeller thrust in kN = displacement in m3

The criterion applies to fully dedicated propellers with suitable skew and tip unloading. The tentative upper limit of 7 is related to the maximum allowable excitation level by the vibratory forces of very large single-screw container ships as a resultant of the pressure fluctuations on the hull above the propeller. It has been found that attaining a favourable wake distribution behind the hull in the very first

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


the optimised sectional profiles for this class of propellers can be properly assessed only by model experiments. By applying the most sophisticated tools, the design of a 25 knot, 90 MW single-screw container ship for about 10,000 TEU would probably just be feasible. A well-balanced, fully dedicated design, in which the characteristics of the main engine, the depth of the wake peak, the tip-hull clearance and the propeller particulars are treated simultaneously from the very first stage of the design, is considered imperative for this class of extreme concepts. Another feature to be looked into in more detail is the risk of rudder erosion due to cavitation. From the growing experience which emerges from these high-powered container ships there appears a clear need to avoid cavitation erosion of the rudder or to keep it under control, if it is not fully avoidable (see section 8.4). A similar feature to pay attention to is the maximum shaft loading and its impact on the maximum possible stern bearing loads. This applies both to the normal ahead sailing conditions and also in particular, to off-design conditions, such as sailing in a turn at high speed and during stopping manoeuvres. Tools to accurately predict the lateral propeller forces and moments leading to the maximum magnitude of the bearing forces in off-design conditions will have to be improved.
2.3. Pod Drives

knots are under development. Most of the pods are installed in cruise liners, but other applications include RoRo ferries, bulkers, supply vessels and icebreakers. Lobachev & Tschitscherine (2001) introduced a RANS method for estimating the model and full scale resistance of a pod with and without a working propeller. For the investigated thruster configuration the full scale resistance coefficient of the unpropelled thruster was 63% of its model value. For the case with the propeller present, the model resistance coefficient of the pod and strut had to be reduced by 31% in order to get the full-scale value. Heinke (2001) and Karafiath (1998) give an overview of the influence of some geometrical particulars such as gondola diameter and length of the gondola on the open water characteristics of pod drives. The relationship between the propeller thrust and the unit thrust for different geometrical configurations is presented. Heinke (2001) describes model test results for a corvette with pulling pod systems and considers the wake field, power requirements, manoeuvrability and cavitation behavior. He concludes that due to effects of cavitation the strut should be asymmetrical. Friesch (2001) shows results of investigations in a large cavitation tunnel. He shows that for twin screw vessels, an advantage can be seen for pulling pod propulsion concerning the vibration and noise behavior compared to conventional propulsion. Friesch (2001), Praefke et al. (2001) and Veikonheimo (2001) describe investigations of a pulling pod arranged as a contra rotating unit behind a single propeller. All come to the conclusion that there is a possible power improvement of about 10% compared to a conventional single screw arrangement. Friesch (2001) identified potential cavitation problems for the down stream pod propeller and strut. The measured pressure amplitudes of the higher harmonics of the blade rate are twice as high as those known for conventional twin screw vessels.

Pod drives are propulsion units which incorporate an electric motor located inside a fully azimuthing pod. They have been implemented on large ships since 1990. Up till now (2001) more than 60 units have been installed, and overall 120 units are on order. These units have a maximum installed power of 21 MW, and are designed for speeds up to 26 knots. Units with 30 MW and a possible speed of 30


The Propulsion Committee

23rd International Towing Tank Conference

Szantyr (2001) presents results of model experiments which included the measurement of drag forces, lift forces, and moments for pod models without a propeller, with a pulling propeller and with a pushing propeller for different yaw angles and advance coefficients. He concluded that the drift angle has a marked effect on the axial force. The axial force for moderate non-zero angles is higher than for zero yaw angles. Mewis (2001, 2001a) presents results of measurements of open water characteristics for different test configurations. The measurements show that the measured propeller thrust is affected significantly by the width of the gap between the propeller hub and the pod housing (Figure 2.7). The measured unit thrust is not affected by the width of this propeller hub gap and less affected by the width of the strut gap (Figure 2.8). Results of pressure measurements inside the propeller gap for different test conditions were investigated. He comes to the conclusion that only the unit thrust (not the propeller thrust) should be used for comparisons of propulsion system efficiency. Mewis (2001) compared the open water efficiency of a pod system with that of a conventional system, using the same propeller. The results show about 5% lower efficiency for the pod unit at the self propulsion advance ratio. Holtrop (2000) has made a distinction between the blade thrust and shaft thrust in treating pod propellers. Knowledge of the drag of the pod housing and strut in the condition when the propeller is working is important. Knowing the drag is essential to assess the scale effect correction to be applied in the extrapolation of the results of model tests on pods. Moreover, most propeller design methods are based on the thrust of the blades. The drag of the pod housing and the strut can be defined experimentally as the difference between the propeller thrust and the unit thrust, the latter being measured at the junction between the hull and the pod strut. The drag of the pod housing is reflected accurately by this thrust difference if the propeller thrust is a consistent entity of

which the level is known without any doubt. However, the thrust exerted by the blades on the hub departs up to several per cent from the shaft thrust due to the shape and the size of the hub and the other particulars of the pod configuration which govern the pressure in the gap between the pod housing and the hub. To assess the blade thrust, which is supposed to be a better entity to be used, a propeller open-water test on the single propeller is used, although propeller open-water tests on pulling propellers with thick conical hubs do not always give satisfactory solutions.

Figure 2.7 Open water characteristics of a pod unit based on measured propeller thrust for different propeller gap widths (Mewis, 2001).

Figure 2.8 Open water characteristics of a pod unit based on measured thrust of the unit for different strut gap widths (Mewis, 2001).

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I



highly desirable to reach concurrence before the final conference. Unfortunately intra Committee review is not a task requirement of Committees, and therefore, if performed, places an additional workload burden on cooperating Committees. The content of the Model-Scale Cavitation Test procedure will not be reviewed in detail in the Committee report. Changes from the previous procedure generally involve detailed procedural descriptions, which can be studied in the updated Quality Manual.
3.2. Description of Cavitation Appearances, ITTC Procedure 7.5-02-03-03.2

In particular, the following procedures should be reviewed: Model Scale Cavitation Pattern Tests ITTC Procedure 4.9-0.3-03-03.1 Description of Cavitation Appearances ITTC Procedure 4.9-03-03-03.2.

3.1. Model-Scale Cavitation Test, ITTC Procedure 7.5-02-03-03.1

The procedure, Model Scale Cavitation Pattern Tests, was renamed and is to be called Model-Scale Cavitation Test. The intent is to broaden the pattern test procedure to be a general cavitation testing procedure including pattern and inception testing. The content of the new procedure is similar to the previous one, with the name change being the most significant apparent modification. The procedure is intended to be a basis for procedures addressing specific areas, such as water quality specification, measurements of the pressure fluctuations, which were being considered by the 23rd ITTC Specialist Committees. In this way, the specialised procedures would not have to restate basic cavitation test procedures, but can reference this procedure instead. By restructuring the cavitation test procedure it became desirable to coordinate its review with not only the QS group, but also the Specialist Committees on Cavitation Induced Pressures, and Water Quality and Cavitation. Feedback from the Specialist Committees took about one year, but was obtained in sufficient time to prepare final procedure drafts to the QS Group. Presently, the review of procedures by related Committees or specialist groups is not a requirement of the QS process, but given the Committee overlap of many procedures it is

The procedure describing cavitation appearances was modified to a limited extent from the previous version prepared by the 22nd ITTC Quality System Group. Primary changes included updating some cavitation descriptions, and including some more recent photographs. Outdated descriptions of cavitation, including spot and field cavitation were removed. Root cavitation was added. Both vortex and bubble cavitation were categorized by specific types. New descriptions are shown below.
root cavitation thick three dimensional cavitation occurring at the blade root, commonly seen on controllable pitch propellers (CPP) vortex cavitation trailing, detached tip vortex cavitation incepts downstream of the blade tip leading edge vortex cavitation occurs along the leading edge, usually at high leading edge loading conditions attached tip vortex cavitation occurs very near the blade tip, often attached to the blade bubble cavitation small bubble type cavitation indicative of propellers with new blade sections with no suction peaks at the leading edge at Re0.7R above about 106


The Propulsion Committee

23rd International Towing Tank Conference

large bubble type, usually isolated, indicative of Re0.7R less than about 106.


further experience is gained. So the Committee sees the procedure that has been written as a first step in this process and not as a final and complete document. The two approaches to extrapolation are described in the form of a check list or summary of the approach as well as in the form of a flow chart to emphasize where similarities and differences exist. The procedure also includes descriptions of how open water propeller tests, open water podded propulsor unit tests and self propulsion tests are done. These sections emphasize the special nature of these tests when applied to podded propulsor systems and concentrate on the differences in the methods from conventional tests. Open water tests on the propellers of these units (without the pod) are done to:

The Committee approached this task to prepare a procedure for predicting the performance of ships fitted with azimuthing thrusters as the main propulsor by reviewing methods currently in use at the major testing tanks. It became clear that use is being made of open water tests of the propellers themselves, open water tests of the podded propulsor unit, self propulsion tests of the ship model fitted with the podded propulsor unit as well as resistance tests of the hull with and without the unpropelled pod and strut. Not all of these tests are used by all experimenters to predict powering performance of the full scale ship. However, each test has been used, or is being used for a certain step in the process, to find detailed information on the podded propulsor unit, or to gain information on the way in which these units work. It also became clear that two distinct main approaches were in regular use for powering prediction: extrapolation using load varying self propulsion tests only and an approach using resistance tests together with open water tests of the podded propulsor system as a unit. The approach taken by the Committee was to write a procedure that included these two main approaches to extrapolation of powering performance. It was also recognized that not all of the details of the corrections to be made in extrapolations for these ships have been determined and so flexibility in the procedure has been allowed to take into account developments of the methods with time and the variations between testing tanks. It is expected that future conferences will refine this procedure as

determine the open water characteristics of the propeller to be used in propulsion tests and for further propulsion predictions; determine data for propeller design, such as wake fraction; estimate characteristics of the propeller for use in detailed design.

An area where particular care has to be taken is for pod units where the propellers have relatively large hubs with extreme conical form. For these propellers the thrust may deviate substantially from the thrust of a propeller with a cylindrical hub. Different approaches are needed to determine the hub geometry and aft fairing pieces used for tests on pushing and pulling type propellers. An example is shown in Figure 4.1. Open water propulsion unit tests are done to: determine the open water characteristics of the whole propulsion unit to be used in propulsion tests and for propulsion prediction; determine data for propulsion prediction, for example the wake fraction;

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


optimize the pod units; assess the effect of Reynolds numbers on performance; compare open water test results of unit and propeller, to gain information on the impact of pod and strut on propeller characteristics.

Aft fairing piece (rotating with hub)

Forward cap


Figure 4.2 Pod-drive in the propulsion unit test (courtesy of HSVA). A procedure for carrying out open water propeller tests, open water unit tests and self propulsion tests for ships fitted with podded propulsors has been written. The procedure includes the description of extrapolation methods for powering performance based on two main methods which may be seen as alternatives. Podded propulsor experience is being developed rapidly and new knowledge is being gained which means that the procedure is seen as a first step in an ongoing process to refine procedures for testing and extrapolation methods for ships fitted with these types of propulsors. The procedure is expected to evolve in the future, and is therefore accepted as an interim procedure. The new procedure is entitled, *Podded Propulsor Tests and Extrapolation, ITTC Recommended Procedure 7.5-02-03-01.3 (*Interim Procedure).

Figure 4.1 Example of hub geometry for an open water test with a tractor type propeller (courtesy of MARIN). These are a relatively new form of tests special to podded units. A diagram of the set up is shown in Figure 4.2. Examples of some of the precautions to be taken with these tests, included in the procedure, are concerns over the width of the gap between the propeller hub and the pod, the width of the strut gap between the top of the strut and the supporting fairing, issues affecting measurement of shaft thrust, unit thrust and torque, etc. The self propulsion test is needed as part of the powering prediction procedure. Concerns, that are included in the procedure, include avoidance of air leakage from the hull through the connection with the pod strut to prevent the propeller becoming ventilated, turbulence tripping issues, drag of the pod housing, optimization of pod orientation, methods of measuring pod unit loads, etc.


The Propulsion Committee

23rd International Towing Tank Conference


mm may not be sufficient to properly define the edge sectional shape. An example of propeller leading edge anomalies documented with measurements demonstrates the problem, see Figure 5.1 from Michael & Jessup (2001). The measured data was obtained from a coordinate measuring machine (CMM) with accuracy on the order of 0.0001 (0.0025 mm).

0.1 X 0.1mm

Cavitation testing of model propellers requires special attention to the geometric accuracy of the propeller geometry. Special consideration of the accuracy of primarily the blade edges is needed to ensure proper simulation of full-scale cavitation performance. The prepared Procedure on Model-Scale Cavitation Tests by the 23rd ITTC Propulsion Committee has proposed minimally defined model accuracy requirements as follows (see ITTC recommended procedures): The geometry of the propeller model is to be inspected prior to testing. This should include a visual inspection for nicks and local damage and subsequent repair. Manufacturing accuracy should be verified to ensure the geometry is within prescribed manufacturing tolerances. For the case of a controllable pitch propeller the selected pitch must be carefully verified. Effort should be made to ensure the propeller model does not deform under test operating conditions beyond what would be expected to occur full-scale. Blade surface global tolerance of 0.05 mm for a typical 250 mm diameter propeller is considered acceptable. Leading edges and tip edges require a higher level of accuracy, which is very difficult to manufacture and inspect. The 23rd ITTC Propulsion Committee has initiated planning of an ITTC recommended procedure addressing this topic. The primary issue relates to the accuracy of the blade edges, viz. the leading, trailing and tip edge geometry. The global accuracy of 0.05

y, mm


Inspection Data Inspection Data Design



x, mm


Figure 5.1 Leading edge propeller measurements compared to design section offsets (Michael & Jessup, 2001). These measurements require a number of manipulations to create the plot as shown. Measurements are performed first on one side of the blade, and then the propeller is flipped over to measure the other side. Referencing is performed from one side of the hub. If the two sides of the hub are not parallel, an error in blade section rake can occur causing a mismatch in the upper and lower surface at the leading edge. By correcting section rake, the offsets from the back and face measurements can be merged. Comparing measured data to section offsets also requires converting the measurements in the propeller coordinate system into the helical system of the blade section. A best fit match of the measured offsets to the design offsets will produce values of measured pitch, rake and skew. The measured section chord length can also be adjusted to obtain a best match to the design section shape. Then deviation of the offsets from the design geometry can be used to compute the measured thickness and camber distributions.

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


In the tip region the geometry definition is very important. Tip vortex cavitation inception is very sensitive to geometrical details. Therefore the entire tip region requires a high geometrical accuracy. A greater density of measured points, both along the chord and span are required in the tip region. An analogy can be drawn to full scale propeller tolerancing and inspections, which often require a specified number of tip edge gauges to be applied. For cavitation quality propeller models, the following considerations are listed below: CMM measurements at specified blade sections, such as 0.5, 0.7, 0.9R with sufficient resolution to define leading, trailing edges and overall section contour; preparation of edge gauges, with inspection processes similar to full scale, but unfortunately, miniaturized; simplified edge gauges using standard radius gauges; with NC-cut blades and minimal hand finishing, assume the geometry is correct, requiring no or little detailed inspection; for global surface inspection of pitch, sparse measurements in chordwise and spanwise directions. As number of measurement points increase, resolution of geometric deviation increases; use of laser sheet leading edge inspection machine for hand working leading edge to projected contour. Systems of this type typically magnify the leading edge contour by a factor of 30 (Anon, 1981); for CP-type model propellers checks should be made of the pitch of the blades after final installation of the blades in the hub. Angular changes in the settings should be properly translated into changes in the axial displacement of positions on the blades, taking into account the radial shifts of the reference points. The central position of the shaft hole, its concentric position, the axi-symmetry of the recesses in which blades are fitted and other geometrical details of the hub should be carefully checked; fillet radii, if specified, and their chordwise distributions along the root, should be checked;

dynamic balancing may be required to reduce vibration during the model experiments; blade numbering, application of marking lines, and application of leading edge roughness should be thin enough to avoid triggering cavitation.

5.2. Improvement in the Prediction of Propeller Backing Performance from Computations and Experiments

The backing condition is one of the most complex and challenging propeller operating conditions to analyze. It also results in the most extreme loading conditions, often determining the strength of the propeller. The previous Propulsion Committee reviewed performance prediction methods for off design conditions including backing. The most advanced analysis method to date was concluded to be a RANS simulation of the blade flow. For the prediction of propeller strength, the determination of the unsteady flow and unsteady blade loading is required. There has been some advancement in the application of unsteady RANS but without validation of computed blade loads. A fully validated analysis method is not expected in the near future. Therefore, presently, prediction of mean and unsteady blade loads will require reliance on experimental procedures. The crashback condition is dominated by the interaction of the free stream flow field with strong recirculation driven by the local propeller induced velocity pushing fluid forward against the incoming free stream flow. Extreme flow unsteadiness directly coupled to varying degrees of blade surface flow separation make computation of forces extremely difficult. Jiang et al. (1996) documented the unsteady ring vortex structure using PIV and related the unsteadiness to the measured unsteady shaft


The Propulsion Committee

23rd International Towing Tank Conference

forces. Jiang identified an oscillation of the ring vortex at a frequency much lower than the propeller rotation rate. Application of CFD to the crashback problem was reviewed in the last Propulsion Committee report primarily with the work of Chen (1996). RANS computations showed large over predictions when compared with open water data which were attributed to effects of cavitation. Viscous effects and extremely large propeller inductions resulted in the tip region of the blade producing attached flow, while the root region exhibited separated flow. Figure 5.2 from Chen (1996) shows the mean flow field in the axial plane, which exhibits a ring vortex outboard of the tip. Chen modeled only a single blade passage, thus assuming blade periodic flow. This assumption precludes simulating the time varying, spatially nonuniform flow field, and can only simulate unsteadiness within a blade passage.

structure. Unfortunately, the computation was not validated with any experimental data. It is well known that RANS methods do not successfully capture the details of unsteady leading edge separation, such as that occurring during dynamic stall of an airfoil. It has been proposed that large eddy simulation (LES) techniques will be required to capture the instantaneous leading edge vortex shedding phenomena likely to occur during crashback conditions.

Figure 5.3 RANS Simulation of crashback flow (Davoudzadeh et al., 1997). Analysis of crashback conditions require steady and unsteady data including measurements of the near-propeller flow field, blade surface flow, net shaft forces, and individual blade forces. LDV measurements can capture the time average flow. The unsteady flow, being nonperiodic, requires instantaneous velocity measurement techniques, such as PIV. Measurements upstream and downstream of the propeller in regions of high flow gradients could be used to derive instantaneous propeller blade section inflows. Measurement of the blade surface flow is most challenging. Figure 5.4 shows nonperiodic cavitation on propeller 4381 tested in crashback at J = -0.7. Over numerous video records, cavitation occurred on different blades

Figure 5.2 Chen (1996) RANS solution, open water. Unsteady RANS was applied to the motion simulation of a submarine in a crashback maneuver, reported by Davoudzadeh et al. (1997). A full simulation of the entire submarine body and rotating propeller was performed during which the propeller rotation rate was reversed. All the propeller blades were modeled separately to capture the unsteady flow field within the entire propeller disk. The flow field was analyzed in the vicinity of the propeller showing the unstable character of the vortex ring structure. Unfortunately, the

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


at different regions of the propeller disk. The leading edge cavitation visualizes the occurrence of unsteady nonperiodic blade separation, which requires measurement of the instantaneous blade flow. The simple technique of cavitation observation can be used to qualitatively describe the unsteady blade loading, but care should be taken to avoid unwanted cavitation when conducting tow tank tests, as was hypothesized by Chen (1996). Instantaneous measurement techniques could be used, such as surface pressure gages, blade local PIV, or blade coatings that visualize shear stress or blade pressure. Figure 5.5 shows an example of instantaneous flow measurements during crashback using PIV. The upper figure represents the time-average flow obtained from averaging 500 PIV images, all obtained at the same blade angular position. The lower figure shows an instantaneous image, selected as a worst case, where flow reversal at the tip is observed. This could be the flow condition at which the cavitation events seen in Figure 5.4 occur.

-60 -50 -40 -30

X [mm]

-20 -10 0 10 20 30 50 60 70 80 90 100 110 120 130 140 150 160 170

-60 -50 -40 -30

X [mm]

-20 -10 0 10 20 30 50 60 70 80 90 100 110 120 130 140 150 160 170 180

Radius [mm]

Figure 5.5 Time-average (top) and instantaneous (bottom) PIV measured flow field about P4381 during crashback condition, J = -0.7 (courtesy of NSWCCD).

Figure 5.4 Nonperiodic leading edge cavitation in crashback condition, P4381 (courtesy of NSWCCD). Another simple technique is the use of tufts attached to the blade surface. Figure 5.6 shows outward radial blade flow at near shaft-stop condition during crashback. Observation of tufts simultaneously over the propeller disk during crashback can qualitatively show regions of separation and preferred flow direction.

Figure 5.6 V=1.9 m/s, n=1.5 RPS, suction side (courtesy of KSRI). Measurement of net unsteady shaft forces would provide total blade forces. Individual unsteady blade load measurements would permit discrimination of individual blade loads on net lateral and axial loads. Blade loads can be measured with in-hub flexures (as shown in


The Propulsion Committee

23rd International Towing Tank Conference

Young & Kinnas, 2000; Moore et al., 2002) or strain gages mounted on individual blades. Comparison of the computations and the measured results requires careful phase measurements, and correlation of the flow field measurements and the various load and/or pressure measurements.
5.3. Developments in the Formulation of Flat Plate Friction

G = 1.0096 + 0.0456 [Log( Rn ) 7.3] 0.013944 [Log( Rn ) 7.3]2 + 0.0019444 [Log( Rn ) 7.3]3 (20 10 6 < Rn < 6 10 9 )

These expressions supersede those of Grigsons 1993 paper presented in the 22nd ITTC Propulsion Committee report. Figure 5.7 compares plots of the various versions of Grigsons lines as well as the ITTC 1957 model/ship correlation line. Shown are the 1993 approximate formulae (Grigson, 1993) which are valid in the ranges of Reynolds numbers from 1.5106 to 2.0107 and from 1.0108 to 4.0109; the revised approximate formulae given at the beginning of this section (Grigson, 1999), as well as an implementation of the direct iterative algorithm. The latter two approaches are also compared in Figure 5.8, which plots the ratio of Grigsons friction coefficient to the values of the ITTC 1957 line (Cf /Cf-ITTC) against Reynolds number. Figure 5.8 emphasizes the approximate nature of Grigsons (1999) curve fits, showing a discontinuity at a Reynolds number of 2.0107 and a sharp rise in friction coefficient at the low end of the valid Reynolds number range which is not present in the direct method (Bose, 2002).

The previous Propulsion Committee of the ITTC monitored the developments in the formulation of flat plate friction. It was realized that through an improvement of the various elements of the extrapolation system a higher accuracy of prediction will gradually be attained. The claimed improvement of Grigsons formulation is based on its theoretical development and the observed improved consistency of form factors in geosim research. However, as outlined earlier a thorough check on a large independent sample of model-ship correlation data is needed to substantiate the application. Before the formulation of Grigson can be promoted to be used in extrapolation it is evident that this independent statistical check is needed to verify that the dispersion in correlation data reduces significantly. In the meantime Grigson (1999) has reworked his approximate numerical representation by which he expressed the results of his algorithm. The following two expressions for G describe the flat plate friction:
C F Grigson = C F ITTC'57 G G = 0.9335 + 0.147 [Log( Rn ) 6.3]2 0.071 [Log ( Rn ) 6.3]3

0,004 friction coefficient

ITTC Grigson 1999 Grigson 1993 Part 1 Grigson 1993 Part 2 Grigson direct



0,001 1,00E+06


1,00E+08 Reynolds number



Figure 5.7 Plot of Grigsons friction lines and the ITTC-1957 model ship correlation line.

(1.5 10 6 < Rn < 20 10 6 ) and

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


Grigson friction coefficient/ ITTC 1957 line




Grigson direct Grigson 1999

0,9 100000


10000000 100000000 Log(Reynolds number)



Practical use of the formulation in extrapolation is still lacking. Through an exercise, form factors of a sample of 513 model resistance tests have been established from lowspeed measurements at MARIN on the basis of Grigsons formulation. Next, the resulting form factors have been related to the ratios of the main dimensions and the hull form parameters by regression analysis to examine the scatter and correlation of the residual error with the Reynolds number. This led to the following prediction formula of 1+k on a basis of Grigson:
1 + k = 1.01 + 0.257059 ( B / LWL )1.34654 (1 C P ) 1.98695 (1 CWP ) 0.475205 C P (TM / B) 0.405704 Exp [0.0190287 CStern 0.0000224964 CStern + 0.468814 (TA TF ) / TM ]
3 2.91432

Figure 5.8 Ratio of Grigsons lines to the ITTC 1957 line. The development of a more easily used version of Grigsons direct algorithm, which involves two iterative steps, might be a more attractive solution for regular use. In the absence of this, the numerical values of the ratio Cf /Cf-ITTC which are plotted in Figure 5.8, are shown in Table 5.1 and can be interpolated. Table 5.1 Tabulated values of the ratio of Grigsons friction line to the ITTC 1957 line (using Grigsons direct method).
Reynolds number 100000 200000 300000 500000 1000000 1500000 2000000 2500000 3000000 5000000 10000000 20000000 50000000 1E+08 2E+08 3.00E+08 5.00E+08 1.00E+09 2.00E+09 4.00E+09 6.00E+09 Cf / Cf-ITTC 0.91143 0.92664 0.92679 0.92454 0.92454 0.92836 0.93305 0.93777 0.94226 0.95725 0.98042 1.00256 1.02532 1.03685 1.04453 1.04780 1.05116 1.05497 1.05843 1.06168 1.06347

CStern is a parameter indicating the afterbody shape:

Shape of afterbody Barge type form with gondola V-type sections Average type of sections (between U and V) U-type afterbody sections CStern -25 to -20 -5 to -10 0 5 to 10

Further, TA is the draught aft, TF is the draught forward, TM is the mean draught, CP is the prismatic coefficient based on the length on the waterline LWL, B is the breadth and CWP is the waterplane area coefficient. The following conclusions were drawn from this exercise: Form factors on Grigsons basis are mostly higher than those on the ITTC-1957 basis. This would be expected from the difference in level of the two formulations at low Reynolds number. The form factors of very small and slender models which are extremely low on the basis of the ITTC-1957 line become of a more realistic level if Grigsons formulation is used as a basis.


The Propulsion Committee

23rd International Towing Tank Conference

Problems to accurately define the form factor due to undulations in the test data, curved lines in the Prohaska plot, sharp deflections of the test data, etc., remained. Extrapolation of propulsion data using Grigsons line with form factor requires lower values of the model-to-ship correlation allowance coefficient CA. Unfortunately, at this stage no conclusions can be drawn yet about the reduction of the scatter of CA, if any. The regression analysis of the experimentally determined form factors did show a slight increase of the standard deviation when compared to a similar analysis of form factors on the basis of the ITTC-1957 line, using the same set of regression variables. This suggests that the natural variation is certainly not reduced when turning to Grigsons friction formulation. However, the form factors determined from the data depend to a certain extent on the skill and judgement of the analyst in charge. A systematic difference could easily have been present here. The regression analysis showed that the Reynolds number dependency of the residual error is absent when Grigson is used as a basis instead of the ITTC-1957 line.

volved for more than 20 years. In his discussion at the 21st ITTC (1996) Ikehata explained that Prohaskas method is not applicable to the ballast condition of most bulbous bow ships. The Propulsion Committee of the 22nd ITTC (1999) reported that the use of a form factor for taking into account the three dimensional flow is not used at all model basins, partially due to the difficulties in determining the form factor from a model test. The difficulties of the form factor method have been summarized by the Powering Performance Committee of the 17th ITTC (1984) as follows: separation on a model may give too high a form factor; laminar flow on a model may give too low a form factor; wave breaking may disturb the linearity of the resistance coefficient; a bulb may also disturb the linearity; interaction between propeller and hull may influence the form effect, an effect not included when the form factor is derived from low-speed resistance measurements, but would be included if self propulsion tests were used for form factor determination (Holtrop, 2001); it may be difficult to take the appendages into account; tank blockage may influence the form factor; the form factor could be dependent on the Froude number; the form factor may be dependent on the Reynolds number. In spite of these difficulties, the Committee concluded that the form factor method is more consistent than the previously used twodimensional method and thus gives rise to less problems in practice and that Prohaskas method is preferred for the following reasons: it is very simple and straightforward to use; it is normally performed by hand and common sense and experience are used to avoid serious mistakes.

5.4. Experimental Determination of Form Factor

In 1978 the ITTC accepted and recommended the 1978 ITTC Performance Prediction Method for single-screw ships to be used. The 1978 ITTC method outlines that the form factor should be determined from measurements of the resistance in the low speed range of 0.12 < Fr < 0.2 using Prohaskas method, in which the resistance is linearized in the form:
CT / C F = ( 1 + k ) + c Fr n / C F

with the exponent n being selected to obtain the best linearization of the data. Since then, there have been many discussions about the consistency of the method and possible alternative methods to determine experimental form factor, but no effective alternative has been proposed by the various ITTC Committees in-

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


The form factor may be determined by a computer using e.g. least square method, but its final choice requires personal judgment. On the other hand, the Powering Performance Committee of the 16th ITTC (1981) suggested several countermeasures to address the practical problems concerning the determination of the form factor. The analysis should be usually limited to the speed range 0.12 < Fr < 0.2, but the upper and lower limits may be extended by about 0.02. However, it may be better to set a lower bound on model speed to account for measurement accuracy, Reynolds number effects, etc. The exponent n should be varied so as to obtain the best least squares fit to the data and corresponding k value derived. The exponent n should be varied so as to obtain the best least squares fit to the data and corresponding k value derived. In 1990 the Powering Performance Committee of the 19th ITTC gave extensive practical guidance in its report how to determine the form factor to be used in the extrapolation of model test results. Reviewing first what had been said by earlier Committees and surveying discussions at previous conferences, the uncertainty of form factor determination in relation to the supposed increased accuracy of the final predictions had been discussed. Treated were the uncertainty of the experimental determination of form factor, effects of surface tension when testing small ship models, effects of tank blockage, effects of Reynolds number, effects of Froude number and the urgent need to have a standardised procedure to improve consistency, accuracy and the uniqueness of the form factor. The 19th ITTC Powering Performance Committee concluded that despite the merits of reducing the dispersion in the correlation, a new factor of uncertainty is introduced. Because form factors may vary considerably between the models of ships of the same main dimensions and the same hull form coefficients, the individual experimental form factors are generally to be preferred. On the other

hand, in many applications statistical formulations of the form factor may be useful. The following practical guidance has been drawn up, summarising Chapter 2.3 of the report of the 19th ITTC Powering Performance Committee, while updating the text on some minor points. This guidance is still considered valid and could eventually become the basis of an ITTC recommended procedure: Hull-form optimisation testing Owing to the natural variation in the determination of the form factor there is a danger of making errors in conclusion about the ranking of alternative designs and about judging the effects of design modifications. It is therefore concluded that initially power savings by design modifications are assessed using a constant form factor. For a more final assessment form factors are to be selected on the basis of the judgement of the type of modification. If the modification is affecting only the wave resistance, as e.g. in most modifications of bulbous bows, the assumption of a constant form factor is of course correct, even though errors in determining the form factors suggest a difference. If the modification is safely judged to be indifferent as regards wave resistance and other forms of Reynolds number independent pressure resistance the difference in form factor can be determined from test points at higher speeds under the approximation that the wave resistance is equal for both variants:

k = R m1 R m 2 SV m C Fm

Well chosen form factors should eventually be selected to make an accurate prediction of the absolute power level. Optimum trim and draught variation tests Similar problems exist here because small changes in draught and trim could render inconsistencies in the results due to the uncertainty in the determination of the form factor. Effects of a bulbous bow and transom immersion may affect the low speed test points and


The Propulsion Committee

23rd International Towing Tank Conference

hamper an accurate form factor determination in some conditions. Because 1+k should vary smoothly in relation to the draught and trim, adjustments of the form factor from a fixed value can be used only when the mutual relation between the form factor and the draught and trim changes have been properly established. Tests in which flow separation is present It has been stressed that in the model experiments flow separation, if any, should be identified and its effects judged as to whether or not there will be a serious impact on the scale effect correction to be applied. The problem seems to be aggravated as at low speeds flow separation effects are generally more severe, rendering the low speed form factor less reliable. Three typical cases are distinguished: Propeller-hull interaction. The Committee recognised a serious departure between the form factor determined from low speed resistance tests and those determined from low-speed propulsion, including load-variation testing as an indication that flow separation may have been present in the experiment with the highest form factor. commonly observed three The dimensional flow separation, twisting of boundary layer flows, on slender forms is not regarded as a feature hampering the use of the low-speed form factor. For bluff forms, on which the pressure drag is very high, the separation drag coefficient may be assumed to be constant with Reynolds number changes and a form factor lower than that derived from the experiment is to be used in the extrapolation of the results in order to avoid a too optimistic power prediction being made. For consistency reasons the form factor is to be that of an equivalent ship of similar fullness but without separation.

the establishment of form factors. By convention, tests on such craft are extrapolated without form factor. For the extrapolation of tests on medium speed ships with a limited transom immersion additional low speed points may be measured specifically for the form factor determination. Alternatively, the low speed test points may be corrected by an empirical formula for the base drag of the immersed transom stern. Confidence is gained when the form factors of the two approaches agree. Ships with streamlined, flow oriented appendages The use of form factors to correct for the scale effect on the drag of appendages is one of the recognised procedures. The determination of form factors of small appended models seems to be hazardous as the flow over appendages may be laminar and a portion of the appendages may be embedded inside the hull boundary layer. Additional measures are probably needed to ensure determination of accurate form factors: the use of not too small models, the use of flow tripping at the appendages and the determination of form factors from tests points at higher speeds under the assumption that the wave resistance is not affected by fitting the appendages. Care is to be taken that the drag of blunt and mis-aligned appendages is not incorporated in the form factor.

Special devices Some classes of energy saving devices are designed to reduce the viscous resistance of the hull by an intentional change of the viscous flow over the hull. Most probably, the best form factor estimate can be made by assuming equal wave resistance at equal Froude numbers at medium speeds and turning to resistance data derived from load-variation tests extrapolated to zero-thrust conditions. Since the devices show a great variability each configuration has to be judged individually.

High speed craft The presence of deeply immersed transoms and a multitude of various appendages hamper

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


5.5. Form Factor Predictions from the Gothenburg 2000 Workshop

State-of-the-art computational fluid dynamic codes for the analysis of ship hydrodynamics have been assessed at the Gothenburg 2000 Workshop (Larsson et al., 2000). Numerical results provided by twenty Organizations were compared with experimental measurements for three selected test-cases: KRISO 300K tanker (KVLCC2): analysis of steady flow without free surface effects and without propeller; KRISO container ship (KCS): analysis of free-surface effects, without and with propeller; DTMB 5415 combatant: analysis of freesurface effects, without propeller. Experimental measurements for KVLCC2 and KCS were provided by the Korean Institute of Ship and Ocean Engineering (KRISO), by Pohang University, Korea, and by the Ship Research Institute of Japan. Experimental data for DTMB 5415 were given by the David Taylor Model Basin, USA, and by Istituto Nazionale per Studi ed Esperienze di Architettura Navale, Italy. Numerical predictions of form factor k = C T / C F 0 -1 are given for the KVLCC2 test case. The selected ship has a typical hull form adopted in modern tankers, with U-shaped stern framelines. The flow pattern is dominated by afterbody inboard rotating bilge vortices near the centerplane and outboard rotating shoulder vortices near the waterplane. Such features yield a difficult task for turbulence models and grid resolution of current CFD codes. Both model scale and full scale conditions are investigated. Test Reynolds numbers are Re= 4.6106 at model scale (scale ratio: 58.0), and Re = 2.03109, at full scale. Experimental values of the total resistance coefficient are, respectively, 4.11310-3, at model scale, and 2.0610-3 at full scale. Equivalent flat plate frictional coefficients by the ITTC 1957 formula are, respectively, CF0,m = 3.4510-3 and CF0,s = 1.40510-3. Resulting form factor values

are k = 0.192 at model scale, and k = 0.466 at full scale (0.25 and 0.39 respectively using Grigsons line). Numerical results obtained by the workshop participants for the KVLCC2 test case are given in Table 5.2 below. Form factor values by using CT from CFD calculations and CF0 from Grigsons line are also shown for comparison and denoted by kGrigson. Computations of viscous flows around ships without free-surface effects were performed by using 2105 to 7.5106 points for the finest grid (average is about 1106 points). Quantitative estimates of grid uncertainties are usually performed. In most cases, results obtained by using no more than four grids are compared. Oscillatory convergence for the pressure resistance coefficient CPR and for the frictional resistance coefficient CF is frequently observed. Some authors note that four levels of grid refinement may be not sufficient to perform an accurate convergence analysis, especially in the case of oscillatory convergence. The ITTC procedure for uncertainty analysis (ITTC Quality Manual Procedure 4.9-04-01-01: Uncertainty analysis in CFD, CFD uncertainty assessment methodology, 1999, 23rd ITTC QS Procedure 7.5-02-01-01) is frequently used. Iterative uncertainty is usually not satisfactorily documented. The statistical analysis of predicted results is helpful to understand the reliability of numerical approaches. In particular, standard deviation and coefficient of variation V = 100/fave, where fave denotes the average among predicted values of a generic quantity f, are considered as a measure of the dispersion of numerical results. At model scale, computed k factors have a standard deviation of 0.066, which represents a coefficient of variation V = 26.4%. At full scale, computed k factors have a standard deviation of 0.136 which represents a coefficient of variation of 31.2%. Complete statistics for CT, CF, CPR and k are summarized in Table 5.3.


The Propulsion Committee

23rd International Towing Tank Conference

It may be observed that the coefficient of variation for CPR is always larger than that of CF. In addition, the coefficient of variation of k at full scale is considerably larger than at model scale. This demonstrates that difficulties are encountered when computing CPR (both at model scale and full scale), whereas reliability of state-of-the-art very high-Reynolds full scale computations is still unsatisfactory. Comparisons of current results with predictions from previous workshops highlight improvements with regards to turbulence modeling, grid generation and high-performance computing. Nevertheless, large differences Table 5.2

among predicted values are still obtained if the same code with different turbulence models is used. Comparisons of point variable results with experiments do not provide clear indications on what turbulence models are preferable. Some authors highlight that improvements in turbulence modelling can be overshadowed by inappropriate numerics; in particular, firstorder upwind discretization schemes for convection terms should be avoided. It is also interesting to observe that different results are obtained by different users of the same codes with the same turbulence models.

Calculations for KVLCC2: top: model scale; bottom: full scale.

CT10 4.392 4.460 4.230 4.700 4.059 4.323 4.090 4.210 4.329 4.340 4.660 3.886 4320

Code Fluent Horus-easm Horus-rij- Horus-sst Fluent Parnassos Neptune Surf CFX UVW CFX WAVIS UNCLE

CF10 3.441 3.630 3.300 3.920 3.357 3.870 3.320 3.370 3.397 2.910 3.590 3.361 3.560

Model scale CPR103 0.951 0.830 0.930 0.780 0.702 0.453 0.772 0.844 0.932 1.430 1.060 0.525 0.760 Full scale CPR103 0.360 0.350 0.320 0.305 0.165

k 0.273 0.293 0.226 0.362 0.177 0.253 0.186 0.220 0.255 0.258 0.351 0.126 0.252

kGrigson 0.341 0.362 0.291 0.435 0.239 0.320 0.249 0.285 0.322 0.325 0.423 0.186 0.319


Code Fluent Horus-easm Horus-rij- Horus-sst Fluent Parnassos Neptune Surf CFX UVW CFX WAVIS UNCLE

CT10 2.110 1.820 2.230 1.766 1.850

CF10 1.750 1.470 1.920 1.461 1.685

k 0.502 0.296 0.588 0.257 0.317

kGrigson 0.416 0.222 0.497 0.185 0.242

1.934 2.236 1.944

1.551 1.940 1.578

0.383 0.296 0.366

0.377 0.592 0.384

0.298 0.501 0.305

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


Table 5.3

Statistics on calculations for KVLCC2: top: model scale; bottom: full scale.
CT10 4.308 0.226 5.2

Average Standard dev. Variation coeff.

CF10 3.464 0.261 7.5

Model scale CP103 0.844 0.243 28.8 Full scale CP103 0.318 0.069 21.7

k 0.248 0.066 26.4

kGrigson 0.315 0.069 21.9

Average Standard dev. Variation coeff.

CT103 1.987 0.182 9.2

CF103 1.669 0.188 11.3

k 0.415 0.130 31.2

kGrigson 0.334 0.122 36.6


Scale effects in open water and propulsion tests are related to model scale Reynolds numbers that differ from full scale by an order of magnitude and more. The dependence of propeller hydrodynamic characteristics on Reynolds number is complicated by differences in friction and in lift characteristics. These scale effects become less predictable because of the variation in the amount of laminar flow in the boundary layer and the occurrence of three-dimensional flow separation from the blade surface. Thus, the basis for extrapolation from model scale to full scale by current procedures is unreliable as only the average extent of laminar-flow and flow separation is accounted for. Attempts have been made to make the basis for extrapolation of propeller characteristics more consistent by turning to tests on leading edge roughened propellers. A firm method for extrapolation on the basis of experiments on propellers with flow tripping has not been developed.

Some aspects related to turbulent flow stimulation on propeller blades are addressed by Bazilevski (2001). In this study, trip wires of 0.1 mm diameter were placed at 10% of blade chord. Propeller thrust and torque coefficients were measured at rotational speeds ranging from 7 to 30 rps which corresponded to Reynolds numbers RnD = nD2/ of 1.2105 to 5105 (model propeller diameter D = 0.14 m) and of 0.5106 to 1.45106 (model propeller diameter D = 0.29 m). Propeller performance values obtained by applying turbulence stimulators on the propeller blades are compared to those obtained with unmodified blades. In the range of rps considered, the scatter of measured efficiency is 13.6%. By using turbulence stimulation, the scatter is reduced to 1.5%. When stimulators are used, measured thrust and torque coefficients are to be corrected to take into account the stimulator drag. Bazilevski (2001) proposes a stimulator drag allowance semi-empirical formula developed by Poustoshny and based on the evaluation of the wire drag according to Tagorys model (Tagory, 1963). Trip wires on the pressure side are not so effective as those fitted on the suction side. This is related to the stability of the laminar boundary layer on the pressure side against Reynolds number variations in the open water test range. In addition, the trip wire effectiveness is largely influenced by the ratio between the wire diameter and the


The Propulsion Committee

23rd International Towing Tank Conference

boundary layer thickness. The choice of the optimal wire diameter should be investigated further. As an alternative to applying scale effect corrections to test results on model propellers which are smooth and which have an uncertain extent of laminar flow, tests could be done on model propellers of which the leading edge is roughened. Due to turbulence tripping the scale effect on the propeller performance becomes larger, but because of the turbulent state of the boundary layer on the model propeller the basis for extrapolation is hypothesized to be more consistent. Boorsma (2000) investigated if model experiments on leadingedge roughened propeller models offer a sounder basis for extrapolation. On the basis of a sample of 5 correlation cases of fixedpitch propellers, of which the relationship between the rotation rate and the power was accurately known at full scale, it was shown by Boorsma that the use of flow-tripping on the model propellers reduced the dispersion of the rotation rate correlation factor for constant power, Cnp, from 2.4 to 1.7 per cent. Further it was concluded that the turbulence tripping was not always effective at the inner radii. The conclusions look promising but the retested sample of correlation cases was too small to justify implementing the procedure as a standard in routine experimental work. Moreover, the change of the procedure requires that the level of the scale effect corrections should be adjusted since the use of leading-edge roughness causes the average levels of the correlation factors to depart from unity. Several experimental investigations of propeller blade flow over a wide range of Reynolds number are available. Some of them are listed by Bazilevski (2001). Flow visualizations on conventional propeller models with diameters between 168 and 355 mm show boundary layer flow which is mainly laminar on both pressure and suction sides below RnD = nD2/ = 1106. At higher RnD, the laminar portion on the suction side is progressively reduced and tends to disappear. On the pres-

sure side, the laminar boundary layer is more stable. Between ReD = 1106 and 1107 a fully developed turbulent boundary layer is established on both sides. The exact ReD value at which this condition is achieved depends on several factors including propeller geometry, load conditions and inflow conditions. In the case of open-water measurements, a convenient procedure is to perform tests at higher rpm than the value required by Froude number identity. This allows operation in a flow regime where laminar flow on the blade suction side, or flow separation at the trailing edge are minimized, and hence closer similarity to full scale conditions is achieved. The procedure above is particularly important in the case of propellers operating in a low-turbulence level inflow, where stable laminar flow at model scale is likely. This is the case for the propellers of pulling thrusters, where the blade flow is only slightly affected by turbulent wakes originated by the hull or by appendages. Open water testing of conventional, large diameter propeller models has documented the threshold above which the propeller performance is independent of Reynolds number. This behavior is related to the development of varying degrees of laminar and turbulent flow, and varying degrees of trailing edge flow separation. Figure 6.1 presents open water coefficients of a conventional propeller, P5370 measured at NSWCCD, operated over a range of Reynolds number at design advance coefficient, J. The thrust and torque coefficients are relatively constant above Re = 1.2106. Qualitatively determined threshold Reynolds numbers Remin, based on inflow at 0.7R chord length, and main geometric characteristics for other propellers are listed in Table 6.1. An interesting trend has been observed when testing both large and small geosim models. Smaller models, which are often used for propulsion tests, have exhibited generally higher efficiency than their larger parent mod-

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


els, which are generally used for cavitation testing, when tested at the same Reynolds number. Shown in Figure 6.1, propeller P5369 (D = 238 mm) run at Re = 6.3105, matched performance of the larger propeller, P5370 (D = 406 mm), at a higher Re of about 1.15106. This trend has been observed in a number of cases, and supports a recommendation of a reduced Re threshold over those indicated in Table 6.1 for small models. The trend observed has no physical explanation. Independent verification would be desirable. The trend may be related to bias errors associated with the detailed testing procedures of the different sized propeller models, blade finishing, and accuracy or differences in blade boundary layer transition properties.
0.8 0.75 0.7

6.2. Scale Effects on the Passive Components of Propulsors

0.65 0.6


KT, Large Prop 1O KQ, Large Prop KT, Small Prop 1O KQ, Small Prop

For complex propulsors, passive components denote those parts of the unit device that interact with the propeller itself. Passive components of propulsors include nozzles of ducted propellers, partial ducts, pre- and postswirl devices, thruster housings and pods. Rudders behind propellers may also be included into this class because they interact with the downstream swirl of the propeller. In addition, in the case of thrusters or podded propulsors, it is important to distinguish between pushing-type and pulling-type configurations. In the former case, the trailing wake shed by the propeller blades does not interact with pod and strut although the upstreaminduced flow field and low-pressure area interacts with the upstream housing and strut to a small extent. On the contrary, in pullingtype configurations, the flow field around pod and strut is largely influenced by the propeller slip stream velocity and static pressure. In such a case, the resistance of the passive components depends on the disc average and radial distribution of propeller loading. Classical scaling laws that are used for conventional propellers are not adequate for complex propulsors. This is essentially due to the interactions between the propeller itself and the unit passive components. The nature of this interaction is not always clear. Similarly, the mechanism by which some unconventional propulsors interact with the hull is still far from fully understood. In particular, it is not known what flow characteristics occur on the full scale and whether these are correctly represented on the model. To obtain reliable extrapolations to full scale of self-propulsion test data, three aspects are of primary importance: accurate scaling of full-scale geometry of test models; accurate scaling of viscous effects; careful consideration of the use of turbulence trips.

KT, 10KQ



Rn x 10-6

Figure 6.1 Open water coefficients of propeller DTMB 5370 (large prop) and 5369 (small prop). Table 6.1 Geometric parameters and Remin for some conventional propellers.
Prop. Diam. T/C0.7R F/C0.7R Skew1.0R Remin106 model (mm) 5343 325 0.0590 0.0219 15.4 1.15 5370 406 0.0464 0.0158 30.0 1.2 5275 406 0.0462 0.0207 45.1 1.25 5168 403 0.0487 0.0332 16.3 1.3


The Propulsion Committee

23rd International Towing Tank Conference

Important effects of passive components are linked to the interaction between these components, the propeller and the hull. These effects are modeled accurately only if the full scale geometry is correctly scaled in test models. Ideally, in powering prediction tests, ship models fitted with unconventional propulsors in many cases should be tested as a unit and not broken down into component tests in which, hull, propulsor and passive components are tested separately. A procedure developed at MARIN to determine scale effects on the drag of the passive components of complex propulsors is described by Holtrop (2001). Specifically, a towing force correction to be applied in model propulsion experiments is derived. Passive components are divided into two classes: nozzle-like bodies and pod-like bodies. First, nozzles of ducted propellers are considered. In this case, viscous scale effects are properly addressed by regarding the interior of the nozzle as a curved plate. Assuming a reference velocity in the axial direction given by nPtip, where Ptip denotes the propeller pitch at the tip, the following expression of the correction to the longitudinal towing force is proposed: 1 F = m nPtip 2 (C Fm C Fs ) C m Dm 2

propeller. A full description of the method is given by Holtrop (2001). The effect of pulling propeller on the drag of the pod housing and support strut can be visualized by inspecting the computed static pressure downstream of a propeller as performed by Chen (2000). This is shown in Figure 6.2. For this configuration, a significant thrust deduction effect would occur if the pod housing support strut were too close to the propeller.

Figure 6.2 Static pressure field about propeller 5168 (Chen, 2000).
6.3. Evaluation of Reynolds Number Scaling Using RANS Computations

where C and D are, respectively, nozzle chord and diameter. For highly loaded ducted propellers the scale effects on the exterior of the nozzle are small. A more complicated analysis is required when dealing with thruster housings or pods. In this case the drag depends on several factors, including the scale factor, the shape of the housing, the orientation with respect to the local flow direction, the interaction with the propeller wake and scale effects on the inflow velocity. A complication is that the local velocity is not known. The housing drag at model scale is defined as the difference between the total unit thrust, and the thrust force that is exerted by the blades on the hub of the

RANS modeling has become a powerful tool to investigate Reynolds number effects on the propulsive parameters. Several workshops have already been dedicated to validation of CFD codes for marine applications (see e.g., the 22nd ITTC workshop on propeller flow, 1998, and the Gothenburg 2000 workshop on ship hydrodynamics). Two aspects have a strong influence on the accuracy of RANS simulations in the range of Reynolds number of interest: grid resolution and turbulence modelling. This is particularly true in the case of viscous flow

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


analysis of both conventional propellers and complex propulsors. Stanier (1998) presents a numerical investigation of marine propeller performance at various Reynolds numbers. Numerical simulations are performed by using a finite volume RANS code based on the artificial compressibility approach. Turbulence modelling is obtained by a three-dimensional version of the Baldwin-Lomax mixing length model. Three propellers are considered: DTRC 4119, DERA C659 and C660. The first is a threebladed unskewed propeller, the others are modern five-bladed skewed controllable pitch propellers. Calculations were performed by using a computational grid with 450000 cells. Numerical results for the DTRC 4119 case at model scale were first compared with experiments in order to validate the approach. At design advance coefficient J, the thrust coefficient was predicted within 0.7% of measurements, whereas the torque coefficient was predicted within 2.5%. Larger errors occurred at lower values of J. Flow field predictions at design J were in good agreement with experiments with the exception of some discrepancies in the tip vortex region that were attributed to grid resolution and to turbulence modelling. Next, numerical simulations at model scale (Re = 0.6106) and at full scale (Re = 3.2107) were compared. In the case of the DTRC 4119 propeller at design J, full scale KT and KQ are, respectively, 2% and 0.3% higher than at model scale. Comparisons of static pressure distributions on the blade suction side reveal differences in the leading edge and tip regions. In addition, full scale results show a larger and stronger tip vortex than at model scale. Larger scale effects are present at low advance design values. A similar trend is observed for the C659 propeller, whereas in the case of the C660 propeller, a 1% reduction of KT at full scale is determined. This is associated with trailing edge flow separation near

the blade tip that is predicted at full scale and not at model scale. The impact of propeller skew on scale effects is further addressed by Stanier (1998a). The numerical investigation highlights scale effects on propellers that depend on several factors including the number of blades, the radial loading distribution, the blade skew distribution. Flow features that are strongly affected by Reynolds number variations include the tip vortex structure and the occurrence of blade flow separation. The theoretical prediction of scale effects of the drag of pulling thrusters is addressed by Lobachev & Tchitcherine (2001). A finite volume RANS code with a turbulence model was used to study the viscous flow around a strut-pod configuration. Propeller effects were accounted for by a body force approach through addition of source terms related to propeller thrust and torque in the momentum equations. The above methodology was applied to the analysis of an azimuthing thruster with and without propeller (dummy model) at model scale (Re = 1.12106) and at full scale (Re = 1.02108). Unpropelled model measurements were used to validate numerical predictions. Total resistance is underestimated with a 9.5% error. Then, model scale and full scale predictions are compared. It is found that full scale resistance is 63% of model scale. In the case of the propelled configuration, the extrapolation factor is 0.69. Thus, the following extrapolation law is proposed: K ship K PSship = K PS model PS K model PS calculated where KPS = DPS/n2D4, and DPS is the pod-strut drag in the propeller-mounted configuration. No attempt is made to determine the sensitivity of the extrapolation factor with respect to J and Re.


The Propulsion Committee

23rd International Towing Tank Conference


the FVM is converted into the effective wake velocity which is used as inflow in the VLM. The results are validated against analytical solutions from actuator disk theory (Choi, 2000). The effect of the grid parameters on the results is found to be very weak. The total velocity field correlates very well with that measured in propeller experiments. Achkinadze & Krasilnikov (1999) claimed that account of radial velocity could be important for propeller design problems. They proposed a reliable and effective iterative algorithm taking radial velocities into account. The examples demonstrated the importance of accounting for the induced radial velocity component in the case of high blade skew and complex rake distribution. Also, they claimed that radial inflow velocity effects are important in the design of propellers for pulling azimuthing pods.
7.3. Development in Boundary Element Methods

Review of the development of numerical design and analysis methods has been a traditional task of the previous Propulsor Committees and the last sessions Propulsion Committee. Further developments in lifting surface and panel methods have been made, often including cavitation. RANS solvers have been further advanced and applied routinely to propellers and podded propulsion configurations. Development of optimization techniques has also advanced, often closely coupled to hydrodynamic analysis, and therefore is included as part of this task.
7.2. Development in Lifting Surface Methods

Lifting surface methods are widely used for the design and analysis of marine propulsors. Since the methods are mature, only a small number of contributions have appeared in the literature recently. Choi & Kinnas (2001) applied the lifting surface vortex-lattice method (VLM), combining a fully three-dimensional Euler solver based on a finite volume approach (FVM), to predict the effective wake for propellers subject to non-axisymmetric inflows. The Euler solver is coupled, through an iterative process, with an existing unsteady cavitating propeller solver. The coupling is achieved in two ways; (a) the pressure distribution computed in the VLM is converted into body forces distributed over the cells which are intersected by the blade surface in the FVM, and (b) the total velocity in front of the propeller computed in

Boundary element methods (or panel methods) are being actively developed and have been extended to unsteady problems for partiallycavitating and super-cavitating propellers. Pyo & Suh (2000) applied a low order potential based panel method for the prediction of the flow around a cavitating propeller in steady and unsteady flow. They used a hyperboloidal panel geometry and a modified split panel in order to improve the solution behavior near the tip region. Usually the trailing edge of the cavity does not coincide with a panel boundary in the chordwise direction for the cavitating propeller, the panel at the trailing edge of the cavity is split into a cavitating part and non-cavitating part. It was found that it is often difficult for the solution to converge near the tip of the propeller since the singularity strengths on the split panels are interpolated from the known values on adjacent panels. They applied a modified split panel method, in which the dipole

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


strength on the non-cavitating part of a split panel is treated as an additional unknown. The method includes a time-step algorithm, i.e., the steady state oscillatory solution is obtained by incremental stepping in the time domain. Figure 7.1 shows computed cavitation extents for a sample propeller operating in nonuniform flow. Dang (2001) adopted a higher order panel method for accurate prediction of the cavity volume and the volume variations in time when lifting bodies travel through a gust or are subject to an ambient pressure change. Calculation was performed to predict steady sheet cavity flows on two-dimensional and three-dimensional hydrofoils and on propeller blades, and good agreement with experimental results was achieved. The dynamics of sheet cavitation was predicted for a hydrofoil moving into a sinusoidal gust and for a propeller rotating in a sharp wake peak. It demonstrated the ability to capture the dynamic movement of the sheet cavitation.

theoretical approach provides reasonable predictions of the most relevant flow features in the propeller wake region. In particular the wake alignment procedure is fully adequate to simulate the trailing vorticity convection process, in which viscosity has a minor influence. Cho et al. (1999) developed a higher order panel method based on B-spline representation for both the geometry and the velocity potential for the solution of the flow around two-dimensional lifting bodies. The selfinfluence functions due to the normal dipole and the source are separated into singular and nonsingular parts. The former is integrated analytically whereas the latter is integrated using Gaussian quadrature (Lee & Kerwin, 2001). Numerical experiments indicate that the method is robust and predicts the pressure distribution around lifting foils with much fewer panels than existing low order panel methods. Ando et al. (1999) developed a simplified surface panel method SQCM, which satisfies the 3-D unsteady Kutta condition. Hess and Smith type source panels are distributed on the hydrofoil and cavity surface. Discrete vortices are distributed on the camber surface according to Lans QCM (Quasi-Continuous vortex lattice Method). Cavity shapes, pressure distributions and forces acting on hydrofoils are calculated for four kinds of hydrofoils. Good agreement is reported between the calculated results and the experimental data, as shown in Figure 7.2. They extended the calculation method for the 3-D unsteady sheet cavitating hydrofoil problem (Ando et al., 2000). Application of the method to the Wagner problem for a rectangular hydrofoil shows the applicability of the method for unsteady calculations. They extended the method to partially and super cavitating cases for 3-D hydrofoils (Ando & Nakatake, 2001). Achkinadze & Krasilnikov (2001a, 2001b) presented a velocity based boundary element method for calculation of partial cavities on

Figure 7.1 Cavity planform at different angular positions on the DTMB 4381 at Js = 0.8, = 2.7 (Pyo and Suh, 2000). In order to assess the accuracy of the solver for the application to modern propellers Esposito et al. (2000) compared the calculated results from a boundary element method for the flow around the propeller of a twin screw vessel with an experimental survey of the flow by LDV measurements. Comparisons with measured velocity fields show that their


The Propulsion Committee

23rd International Towing Tank Conference

wings and propeller blades. A special algorithm of the Modified Trailing Edge (MTE) is used for definition of the lifting part of the flow. The use of curvilinear source boundary elements and continuous distribution of the vortices on the mean blade surface are the distinctive features of the developed algorithm. Prediction of the cavity patterns is performed using an Iterative Cavity Alignment (ICA) procedure with Free Cavity Length (FCL).

the wake deformation also has a strong influence on the resolution of the predicted downstream velocities near the blade tip and root regions. Under moderate load conditions, the wake arrangement has a slight influence on the prediction of thrust and torque coefficients.

Figure 7.3 Comparison of the computed cavity extents (lower) with visual observations (upper) on a podded propeller behind a hull. (Achkinadze et al., 2001). Salvatore & Esposito (2001) presented a theoretical analysis of three-dimensional cavitating flows based on a boundary element approach which includes rotational and viscous flow effects. Partial sheet cavitation is studied by a closed-cavity nonlinear model which includes the prediction of the cavity detachment point. The trailing vorticity path is described by a wake-alignment technique as shown in Figure 7.4. Viscosity effects are included via a viscous/inviscid technique based on a boundary-layer assumption.

Figure 7.2 Lift and drag coefficients versus cavitation number (Elliptic hydrofoil, NACA0012, = 10) (Ando et al., 1999). Achkinadze et al. (2001) extended the high order velocity based BEM for the quasisteady analysis of marine propellers in the non-uniform wake field with comparisons shown in Figure 7.3. The developed method was applied to a multi-stage podded propulsor problem. Liu & Colbourne (2001) formulated a wake model for a time domain propeller panel method covering the near field to far field wake regions as a means of improving downstream field velocity predictions. Results showed that the shape of vortices due to influences of induced velocity components have a strong influence on the downstream flow field velocity prediction because of the orientation of the doublets on the wake panels. Inclusion of induced tangential and radial velocities for

Figure 7.4 Predicted attached cavity and trailing wake (Salvatore & Esposito, 2001).

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


Lee & Kinnas (2001) presented a boundary element method for the numerical modelling of unsteady blade sheets and developed tip vortex cavitation on propellers operating in a non-axisymmetric flow-field. The wake surface is calculated in an iterative way by aligning the surface to the flow velocity. Once the aligned wake surface is determined, the shape of the blade sheet and tip vortex cavity, having constant pressure distribution, is determined by applying the dynamic and kinematic boundary conditions on the cavity surface. The method is applied in the case of a simplified 2-D vortex cavity, 3-D elliptic wing, and propeller blades subject to inclined and nonaxisymmetric inflows.
7.4. Development of RANS Solvers for Open and Ducted Propellers

for the DTRC 4119 propeller are presented. A grid dependence study was performed by using three computational grids with 1.6106, 1.9106 and 2.3106 cells. Velocity fields at a downstream plane are shown. The agreement with experiments is good except in the tip wake region, and this is attributed to unsatisfactory grid resolution in that region. Computed thrust and torque coefficients are plotted against experimental results and numerical results obtained by an inviscid Euler-flow code. The thrust coefficient is in good agreement with experiments, whereas the torque coefficient is underestimated. Preliminary results for a modern propeller with large pitch and skew variations are also shown. Sanchez-Caja et al. (2000) present the application of the FINFLO solver for the analysis of the flow around a ducted propeller. The FINFLO solver is a multiblock multigridstructured finite volume RANS code, initially developed at Helsinki University of Technology for the analysis of compressible flows (Siikonen et al., 1990) and successively extended to incompressible flows by using the pseudo-compressibility technique. The turbulence model is based on a low Reynolds formulation. The geometry of the ducted propeller is the same as that used by Kawakita (1992) where LDV measured velocity fields were obtained at three different advance coefficients. Numerical solutions were obtained using one computational grid with 1.2106 cells for both duct and propeller. The convergence history of the computations was shown for forces, but not for residuals. Predicted and measured forces are in good agreement. At advance coefficient, J = 0.5, the error for the predicted thrust coefficient is less than 1%, whereas the torque coefficient is overpredicted by 4.5%. At different values of J, thrust coefficient errors are slightly higher, and torque prediction is improved. Comparisons for the velocity field at two planes behind the propulsor are also shown. At x/R = 0.65, just behind the duct, the agreement of predicted velocity fields with experiments is good. Fur-

In the last few years, improvements in numerical modelling of viscous flows around marine propulsors have been obtained. In particular, Reynolds Averaged Navier-Stokes (RANS) equation solvers are widely used to analyse viscosity effects on propulsors. Current applications include open and ducted propellers, azimuthing thrusters and podded drives. Flowfield predictions are performed at model-scale Reynolds number in the case of uniform inflow and non-cavitating conditions. Extensions to non-uniform flow simulations and to cavitating-flow analyses are at an early stage of development and practical applications are still not available. In Arnone et al. (2000) a Navier-Stokes code originally developed for turbomachinery computations is applied to the simulation of the flow around marine propellers in uniform inflow. The solver is a cell-centered finite volume RANS code with Chorins artificial compressibility to address incompressible flows. Some different turbulence models are implemented: marine propeller flows were studied by using the Baldwin and Lomax twolayer mixing length model. Numerical results


The Propulsion Committee

23rd International Towing Tank Conference

ther downstream, at x/R = 1.0, the agreement degrades. The authors explain this as a consequence of numerical dissipation. Validation of the FINFLO code for the analysis of an isolated propeller in uniform inflow is presented by Sanchez-Caja (1998), in which the DTMB 4119 propeller is studied. More recently, Sanchez-Caja et al. (2001) considered a four-bladed propeller from the BB series, with expanded area ratio of 0.614 and pitch diameter ratio of 1.046 at r/R = 0.7. One computational grid with 1.3106 cells was used. Axial and tangential velocity fields in the wake region at four locations downstream of the propeller for an advance coefficient, J = 0.8 are presented. Numerical results are qualitatively related to flowfield trends by momentum theory to demonstrate that the methodology is able to capture basic flow phenomena in the propeller wake region.
7.5. CFD Techniques for Azimuthing Thrusters and Podded Drives

The proposed methodology is applied to a thruster consisting of a BB-series propeller with pod and strut. By using the quasi-steady approach, the flow unsteadiness is not modelled and only mean quantities are determined. Computed mean thrust and efficiency differ from experiments by about 8.5% and 6.5% respectively. If the time-accurate approach is employed, the above errors reduce to 0.9% and 0.8% respectively. The timeaccurate approach provides an accurate description of the vorticity shed by the propeller blades and of the instantaneous pressure distribution over both propeller and strut surfaces. The main advantage of the quasi-steady approach is that computational time is reduced by a factor 10 compared to timeaccurate calculations. The coupling between moving and fixed parts is obtained by using a sliding mesh technique. Two different computations are performed around the pod geometry: in the former a quasi-steady approach is followed and circumferentially averaged velocities are used as inflow on the pod, while in the latter a fully unsteady simulation is undertaken.

The FINFLO code referenced in Sec. 7.4 was applied to the simulation of the unsteady flow around a tractor thruster by SanchezCaja et al. (1999). The sliding mesh technique was used to analyse the interactions between moving and fixed parts. Two grid blocks were employed: one for the stationary components (pod and strut) and another for the rotating component (propeller). Since the grid lines across the sliding surface are not continuous, the information is transferred across the surface via suitable interpolation. Two different procedures are used to describe the unsteady nature of the flow: (a) quasi-steady approach, by which a circumferentially-averaged flow is transferred through the sliding surface, and (b) time-accurate approach, where full-unsteady RANS equations for both fixed and rotating parts are solved.

Figure 7.5 Grid on the thruster surface (total number of cells in the mesh is 814080, Sanchez-Caja et al., 1999). The CFD techniques used in the development of a new podded propulsor are described by Vartdal et al. (1999). A mixed potentialflow/RANS solver approach is applied in the design cycle of thruster housing geometry.

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


The propeller flow is solved by a lifting surface code and the averaged velocities are used to assign the inflow in the analysis of the flow around the pod which is performed by a RANS solver. Abdel-Maksoud & Heinke (1999) apply a commercial RANS code based on a finite volume method to the simulation of the flow around a thruster with a ducted propeller and of the flow around an electric podded drive. Both simulations require a complex multiblock grid generation with up to 100 blocks and 4.6 to 6.3105 cells. Some flowfield characteristics are shown for the two cases. Results show many features of the interaction between the housing and the propeller. Lobachev & Tchitcherine (2001) present a numerical investigation of an azimuthing thruster viscous flow in which the emphasis is on full-scale resistance estimation from model-scale results as referred to in sections 2.3 and 6.3.

provided the axisymmetric input for solving the rotor blade flow. The circumferential average velocity, pressure and turbulence were specified as the inflow boundary condition. The rotor gap was not modeled. The computed local rotor flow field was shown. Some error was observed on matching the mass flow from the two solutions. The 22nd ITTC Propulsion Committee already reported some applications of hydrofoil and propeller optimization problems. During the term of the 23rd ITTC Propulsion Committee many other papers appeared. In recent years there is a continuously increasing interest in optimization procedures for marine propulsor design. Classical optimization methods are widely used with improving capabilities. A recent application to the design of counter-rotating propellers and podded propulsors with tandem bow/stern propellers is presented by Achkinadze et al. (2000). Complex propulsors are optimized via an iterative procedure. Each propeller is studied as isolated systems and immersed in the velocity field induced by other components (adjacent stage, blade row or pod) and by the hull wake. The optimum spanwise circulation distribution is obtained from the analytical Generalized Optimum Condition by Achkinadze (1989), which takes into account both axial and tangential components of the radially nonuniform velocity field. A non-linear liftingsurface model is used to define the blade pitch and camber (Achkinadze & Krasilnikov, 1999). This model accounts for radial components of the inflow and self-induced velocity field. The latter is important when designing the highly skewed and raked propellers. The optimum spanwise distributions of blade chord length and maximum thickness are searched via implementation of Lavrentievs optimization method (Achkinadze et al., 2000) using both cavitation and strength criteria. In addition, propeller diameter is optimized in terms of propulsor efficiency. The performance of the designed propulsor is verified by a velocity-based boundary element

Figure 7.6 Overall view of geometry and surface grid. (Paterson et al., 1999). Paterson et al. (1999) applied the unsteady Reynolds-Averaged Navier-Stokes (RANS) solver, CFDSHIP-IOWA, to obtain steady flow solutions for the flow field through an integrated marine propulsor, the MIT Sirenian Propulsor. Steady flow solutions were initially presented for the propulsor modelling the hull, stator blades, and duct, with the rotor modeled as a body force. This solution then


The Propulsion Committee


23rd International Towing Tank Conference

method (BEM) developed by Achkinadze & Krasilnikov (2001). The main features of modern design methods for supercavitating propellers and the history of their development are described in detail by Tulin (2001), Achkinadze (2001a) and Kinnas (2001). Toyoda et al. (1999) developed a new method to design a super cavitating foil section automatically, by applying an optimization method to super-cavitating foil section design. From the 150 foils designed as a series of foil sections, polynomial expressions are obtained by a least-squares method for foil shapes and foil performance such as lift/drag ratio. By using these polynomials, the designer can obtain the foil shape and estimate hydrodynamic performance for a set of a few design parameters including design lift coefficient and strength requirement. A trans-cavitating propeller adopts supercavitating blade sections near the tip and noncavitating sections near the root. Designs using these series of foil sections attain improved performance over conventional super cavitating designs (Kudo et al., 1999). Eight propellers were designed and tested in a cavitation tunnel. The results show that the present design method achieves accuracy of thrust within 5% of experiment. Zondervan & Holtrop (2000) describe a numerical method that is used on a routine basis for the design of blade or strut sections. The method uses parametric representations of thickness and camber distributions. The cavitation inception bucket is evaluated by a two-dimensional panel method. The optimization strategy is based on a random search technique. Figure 7.7 shows results of the optimisation for a family of sectional profiles shaft struts. Four profiles, indicated by I, II, III and IV, were subsequently determined for four speeds: 25, 30, 35 and 40 knots. Corresponding design cavitation indices are 1.17, 0.81, 0.60 and 0.46, respectively. For reference purposes, characteristics of the NACA 0020 and the EPH profiles are also shown. Profile shapes are depicted in Figure 7.8.

IV III 2.0



- C P min

NACA 0020

= 1.17




= 0.81 = 0.60 = 0.46


0 0 2


Figure 7.7 Characteristics of optimized strut sectional profiles and comparison with two conventional two-dimensional profiles (Zondervan and Holtrop, 2000).
0.5 0.4 0.3 0.2 0.1


0 0 -0.1 -0.2 -0.3 -0.4 -0.5 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

x/c I II III


Figure 7.8 Shapes of optimal strut sectional profiles (Zondervan & Holtrop, 2000). The infinite dimensional optimization technique based on the Hilbert space theory is proposed by Jang et al. (2001) for optimizing marine propeller and hull forms. Propeller thrust and torque are assumed to be functionals of the blade pitch distribution. The pitch distribution is expressed as a linear combination of suitable mathematical functions, ei. The optimal distribution is obtained by solv-

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


ing a variational problem which is numerically recast in a finite-dimensional linear matrix system where the unknowns are the weights of the ei functions. Constraints may be included in defining the optimization procedure. In particular, a condition that maximal local pitch changes do not exceed 20% is used. The propeller hydrodynamic forces are obtained under inviscid, non-cavitating flow assumptions, by means of a lifting surface approach. A simple viscosity correction is included to correct the propeller torque. As an example, the MAU type propeller with constant pitch is used as an initial guess. As a result of the optimization process, the efficiency at a given value of the advance ratio is increased by 4%. It may be proved that the optimized shape is unique and is independent of the initial guess used. For a propeller-stator combination, Guner & Atlar (1999) have developed a design method using a lifting-line model. The modelling of the trailing vortex system is made as realistic as possible by taking into account slipstream contraction and roll-up. The stator downstream of the propeller is considered to be a retrofit and is optimized, but simultaneous optimization of the whole configuration was not carried through. Flow conditions at the propeller and the stator are mutually matched. By comparison with other calculations and experimental data the authors concluded that the modelling of the propeller wake is important to give a sound design and a realistic prediction of the benefits of the stator. In addition to applications of conventional optimization methods, a large amount of research has been conducted on genetic algorithms (GA). The main advantage of genetic algorithms compared to conventional optimization techniques is related to the ease of implementation of such methods. In fact, GAs do not require the evaluation of the objective function derivatives. Furthermore, GAs possess good exploration and exploitation characteristics. This means that such methods can be used to investigate a wide region of the deci-

sion variable space (exploration) and, at the same time, are able to obtain the best solution (exploitation). The main disadvantages of GAs as compared with conventional techniques are the large number of steps required in the optimization process and the impossibility to determine a solution which represents the optimum in a global sense. A simple application of GAs is proposed by Wang et al. (2000), where the goal is to design new blade sections to achieve cavitation free buckets that are as wide as possible. Section camber and thickness distributions are given by analytical representations in terms of a limited number of design scalar parameters. The target of the optimization process consists of determining the set of parameters that corresponds to the widest cavitation-free bucket. No constraints are explicitly considered. The GA-based optimization procedure includes crossover, mutation and catastrophe operators in order to accelerate the optimum search and to prevent premature convergence. The pressure envelope is based on the Cpmin value criterion. A Boundary Element Method (BEM) for non-cavitating potential flow is used to determine the pressure distribution on each geometry.

Figure 7.9 Pareto optimal set in the objective function space (f1, f2) and representative optimized shapes. (Esposito & Salvatore, 2000).


The Propulsion Committee

23rd International Towing Tank Conference

A different approach was used by Esposito & Salvatore (2000), incorporating a multiobjective optimization technique for cavitating blade sections. The goal is to minimize cavity length and foil drag for two different values of the angle of attack. Multi-objective optimization is based on the concept of Pareto domination. Specifically, the optimum is searched among all the feasible solutions that belong to the Pareto optimal set. This set collects all the feasible non-dominated solutions, i.e., those solutions that cannot be improved with respect to one objective function without degrading at least one of the other objective functions. In order to reduce the number of control variables, the hydrofoil geometry is described by B-Spline curves for distinct mean-line and thickness. The cavitating flow solver is based on a BEM formulation with viscous effects described by a boundary layer model.

(C) Cavitation of the propeller-rudder system. The effect of cavitation on propulsive characteristics and the risk of cavitation-induced rudder erosion. Investigations into problem (A) have used both computational and experimental methods, whereas problem (B) was investigated using various computational methods. Problem (C), the effect of cavitation, has been studied experimentally.
8.2. Investigations of Propulsion Efficiency

Voigt (1934), Lindgren (1955), Suhrbier (1974), Taljanova & Stoomph (1976) and Stierman (1989) are recognised as the primary publications on experimental investigations of propeller-rudder interactions. There are two experimental approaches available in the study of propeller-rudder interaction: propulsion tests of ship models and open-water tests of isolated propeller plus rudder systems. Stierman (1989), in his experimental study on an isolated system, has made the following observations about the propeller-rudder interaction. There is a longitudinal rudder force, KR, induced by the operation of the propeller. The rudder force is positive, producing thrust at low J values, and negative, producing drag at high J values. This induced force depends on the clearance between the propeller and the rudder, on the relative thickness of the rudder and on the propeller pitch. The rudderinduced thrust is 23% of the propeller thrust even under bollard pull conditions whereas under the design condition it is virtually zero or slightly negative, as shown in Figure 8.1. The presence of the rudder shifts KT and KQ curves somewhat closer towards higher J values: changes in KT at fixed J are 0.010.02 and changes in KQ are 0.0020.004. These values are slightly dependent on J and tend to

One can identify three research areas associated with propeller-rudder interaction developed over the seven-decades-long history of the problem: (A) The effect of the rudder upon propulsion. The component-wise contribution of the propeller and rudder when considered as a combined system for both the behind case and in open water. (B) Computational investigations of steady and periodic rudder force and torque components induced by the operating propeller at zero and non-zero rudder angles.

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


decrease with the growth of the propeller/rudder clearance and to increase with increases in the propeller pitch and in the rudder relative thickness (within 0.01 for KT and 0.001 for KQ).
8 6
KT 10KQ 10KR 10

P/D=1.00 TH/D=0.18 DIST/D=0.30


4 2 0

-2 -4 -6 0,0 0,1 0,2 0,3 0,4 0,5 0,6 0,7 KR

ship model propulsion tests. It was experimentally demonstrated that propeller-rudder interaction patterns in open water and behind the hull were very similar, and the uncertainty of analysis based on open-water tests might be 0.01, especially for the thrust deduction coefficient. It was also demonstrated that reducing the propeller/rudder clearance could raise the propulsive efficiency by up to 1.5%. The researcher has also postulated the significance of the rudder leading edge angle with respect of the vertical plane, with higher efficiency resulting from a vertical orientation of the leading edge.
8.3. Computational Developments






Figure 8.1 KT, KQ values and rudder thrust KR versus advance ratio J (x-axis) Note: y-axis, 10KQ should be 10KQ for clarity. In order to raise the propulsive performance the rudder should be made as thick as possible. This, however, disagrees with data presented by Voigt (1934), Lindgren (1955) and Taljanova & Stoomph (1976) who have demonstrated that there is a propulsion-wise optimum rudder thickness of 2022% of the chord. Particular attention should be paid to the impact of the rudder upon the propeller-hull interaction coefficients. Stierman showed a significant effect of up to 0.030.04 for thrust deduction and 0.06 for wake coefficients, depending on rudder thickness and propeller/rudder clearance. Results were obtained for two ships: a twin-screw passenger liner and single-screw chemical tanker. The presence of the rudder led to 0.010.02 gains in the propulsive coefficient. Investigating the efficiency component due to axial, tangential and viscous energy losses, Stierman has shown that the presence of the rudder reduces tangential losses of the propeller slipstream. Vasiljev (1996) has combined open water propeller plus rudder test results with those of

Earliest developments on analytical methods for propeller-rudder interaction were reported by Tsakonas et al. (1968, 1975). Tsakonas et al. (1968) suggested and then Tsakonas et al. (1975) elaborated a procedure to account for the rudder thickness. Principal details of that procedure can be summarised as follows: arbitrary geometry propeller and rudder are simulated by lifting surfaces interacting with a non-uniform velocity wake; propeller blade and rudder thickness are approximated by a source-sink array. The iterative procedure described in that paper enabled calculations of steady and periodical forces and torque on the propeller, the lateral force and the rudder torque. It was shown that accounting for propeller blade and rudder thickness led to improvement of the procedure. Averaged and periodical rudder forces depended primarily on thickness, and periodic force components depended strongly on propeller-rudder clearance. Those forces were asymmetrical with respect to the blade angle, , and the extent of the asymmetry also depended on the clearance, x0, in relation to the propeller tip radius, r0. Table 8.1 shows a comparison of estimated rudder thrust at the propeller thrust coefficient of 0.37.


The Propulsion Committee

23rd International Towing Tank Conference

Table 8.1 Estimated rudder thrust at the propeller thrust coefficient of 0.37.
K TR = T R / n 2 D 4
(rad) x0 = 0.84 r0
-0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.0646 0.0292 0.0075 0 0.0069 0.0280 0.0629 x0 = 0.72 r0 0.0646 0.0291 0.0074 0 0.0070 0.0283 0.0634 x0 = 0.56 r0 0.0646 0.0288 0.0072 0 0.0073 0.0290 0.0645

tions and propeller and rudder interaction forces was good. It was shown that to get a more accurate solution one should use a doublet distribution on the rudder surface. They also analyzed the case of a propeller in nonuniform flow, simulating the hull wake and a rudder positioned behind it. The propeller thrust and torque and the rudder drag could be accurately estimated within simplified propeller theory and thick-wing theory including a 2D boundary layer simulation. Li & Dyne (1995) developed a linear method under which the propeller free vortices have a pitch and a radius constant in the axial direction, i.e. they are assumed to be undisturbed by the rudder. A similar assumption had been used earlier by Betz (1938), Isay (1965), Yamazaki (1968, 1985), Nakatake et al. (1978, 1981), Moriyama & Yamazaki (1981, 1981a), Tamashima et al. (1992) and Turnock (1993). However, Li & Dyne (1995) focused primarily on the investigation of propulsion, more specifically, the different roles of axial force components that had not been covered by other authors. A non-linear method was developed based on Li (1994). There were four axial force components specified: rudder thrust caused by propellerinduced tangential velocities in the slipstream due to the rudder attack angle in the flow; rudder viscous drag increase due to the fact that the axial velocity in the slipstream is higher than in the uniform flow; pressure drag of the non-zero thickness rudder due to the fact that the axial flow is accelerating just behind the propeller, and the fore part of the rudder is therefore located in a region with a higher static pressure than at the aft part; propeller thrust and torque increase because the rudder of a certain thickness placed just behind the propeller blocks the slipstream of the propeller (because the advance velocity, VA, decreases).

Moriyama & Yamazaki (1981, 1981a) investigated the influence of rudder placed in the propeller slip stream upon the propulsive characteristics. The problem of propellerrudder interaction was theoretically considered with the help of a rudder theory based on wing theory and the consideration of the interaction forces. Surface pressures were measured in the SRI Cavitation Tunnel. Formulae were offered for the pressure on the rudder in non-potential flow behind the propeller, and a method was suggested for estimating rudder resistance with the help of the surface pressure formula and boundary layer theory. Yamazaki et al. (1985) theoretically considered the problem of propeller-hullrudder interaction. They simulated the propeller by an infinite-bladed model, the rudder as a finite-thickness rectangular wing, and estimated individual wakes due to the hull, the propeller and the rudder, and then calculated the performance characteristics. Propeller induced velocity was represented by a system of vortices (bound vortex and free vortex distributed along the helical surface). The rudder was simulated by a bound normal doublet, sources on the rudder surface and a free normal doublet in the wake plane. The frictional drag was found by laminar, transition and turbulent boundary layer calculations. The interactive flow features for both the propeller and the rudder had to be found from the propeller rudder interaction problem solved iteratively. The agreement between computed and experimental results of rudder pressure distributions and propeller

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


Another issue that attracted Li & Dyne (1995) was the amount of the slipstream rotational energy recovered by the rudder and the rudder influence upon the propulsive efficiency under different loading conditions. Propeller performance characteristics were computed from the lifting-line method based on Goldsteins factors. This method calculates the propeller blade circulation, the thrust coefficient and the torque coefficient including the influence of the rudder induced velocities. The rudder is simulated by a set of elements, each represented by a bound vortex filament, a source filament and two free vortex filaments trailing straight downstream. Their locations and the element control points follow the standard for vortex line methods. The viscous drag is estimated as
2 C RD = (1 + 2.4 t / c )C f + 0.234 m

PR =


It was shown that the presence of the rudder behind the propeller slightly improved the open-water efficiency. The highest gain in the open-water efficiency was about 1.7% at J = 0.2; P/D = 1.0 whereas at J = 0.5 PR was 1% higher than 0. One could as well improve the open-water efficiency by increasing the clearance between the propeller and the rudder. Thin rudders seemed to be more preferable than thick ones. The authors concluded that the linear method was a quick and simple tool for propeller-rudder interaction analysis, though not very accurate because it ignored the non-linear effects. Nevertheless, it could be used to obtain basic information on propeller-rudder interaction, to perform parametric studies and to get initial input data for computations with more comprehensive methods. Sding (1998) analysed different aspects of propeller-rudder interaction using panel methods. He emphasised that the potential theory did not model the blade viscous wakes in the propeller slipstream. Nevertheless, Sding has assumed no vorticity for the rudder-induced velocity. He suggested a correction to the resistance increase assuming that the rudder does not change the slipstream velocity, but the slipstream direction follows the rudder midsection plane causing a smaller longitudinal component of the propeller thrust. Another assumption involved is that the propeller slipstream follows the rudder mid-section line. Sding (1998) compared calculations of rudder forces with the developed panel methods and RANS-code against experimental results from Whicker & Fehlner (1958). In Figure 8.3 the comparison of rudder force and torque calculations and measurements show good agreement.

where m is the chord-wise averaged angle of attack. Computational results were compared with experimental data obtained by Stierman (1989) in his investigations into the influence of rudder parameters upon ship propulsive characteristics. Studies also covered the influence imposed on KT, KQ, CXR by a number of parameters: the clearance between the propeller and the rudder, the rudder span length and the rudder chord length. Results of calculations performed by Li & Dyne (1995) are shown in Figure 8.2. The Figure also shows that increasing the pitch produced the greatest impact upon CXR due to the rise in the induced velocities, particularly the growth of thrust at low J.
CXR is defined as C XR = XR

n2 D4

The researchers have also considered different aspects of the propeller-rudder system efficiency calculated as


The Propulsion Committee

23rd International Towing Tank Conference

Ghassemi & Allievi (1999) developed a computational method to estimate hydrody-

namic performance and flow fields around conventional and podded propulsion systems.

CXR versus J at different T/C

KT versus J at different T/C

CXR versus J at different P/D

KT versus J at different P/D

Figure 8.2 CXR and KT versus J at different P/D and T/C (Li & Dyne, 1995). The method employs a vortex-based lifting theory for the propeller and a potential surface panel method for the steering system. Computation of velocity components demonstrated good agreement with experimental measurements behind the conventional propeller with

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


and without the rudder. Calculated thrust, torque and lift also showed good agreement with experiments. The results of application of the method to podded propulsion system, validated by the experiments of Tamashima et al. (1993) showed that in the range of rudder angle 0-20 degrees propeller total thrust and torque are practically independent of rudder angle, in the range 20-30 degrees the thrust reduced continuously to 80% of initial value and torque increased approximately to 115%. At a rudder angle of more than 30 degrees the authors concluded that the computation model was insufficiently accurate.

8.4. Cavitation Aspects

The effect of cavitation on propellerrudder interaction becomes important on high speed vessels, including container ships (see Figure 8.4, Figure 8.8). Cavitation performance will vary with rudder type. Rudders with fixed horns will tend to have higher cavitation inception speeds than spade rudders designed for high speed ships.

transverse force [kN]

1000 500 0 -500 -1000 -1500 -50 -40 -30 -20 -10 0
600 400 200 0 -50 -40 -30 -20 -10 0

attack angle, degree

10 20 30 40 50

Figure 8.4 Interaction of tip vortex with rudder (courtesy of MARIN). Cavitation problems for rudders were reviewed by Kracht (1995). He noted that the rudder in the slipstream of a highly loaded propeller is very susceptible to cavitation erosion at propeller loads in excess of 700 kW/m2 at ship speed greater than 22 knots. Cavitation on the rudder is caused by the local pressure at the rudder blade and by cavitation generated by the propeller convecting into the rudder. Depending on the type and position of the rudder, different zones of expected cavitation attack were identified, categorized into two groups: that resulting from propeller induced cavitation and that stimulated by the flow conditions at the rudder, called selfcavitation. The rudder is affected by the cavitation bubbles convecting from the propeller and collapsing at the rudder surface and on the other hand, by the local angle of attack due to propeller induced velocities which occur even at zero helm angle. The cavitating hub and tip vortices result in rudder erosion along the vortex trajectories as shown in

longitudinal force [kN]

attack angle, degree

10 20 30 40 50

stock moment [kNm]

200 0 -200 -400 -50 -40 -30 -20 -10 0

attack angle, degree

10 20 30 40 50

Figure 8.3 Comparison of Rudder Forces and Moments (Sding, 1998).


The Propulsion Committee

23rd International Towing Tank Conference

Figure 8.5. Kracht emphasized, that the first measure against premature rudder erosion is the choice of non-cavitating profiles which do not minimize bubble cavitation. The profiles should have a chordwise negative pressure distribution without peaks which steadily decrease down to a flat minimum and steadily increase up to zero. So, rudder profiles should be developed to minimize cavitation for the rudders of a high speed ship.

outboard propeller rotation erosion effects were observed only on the outboard surface. Shen et al. (1997a) carried out tests in the Large Cavitation Tunnel at the Naval Surface Warfare Center. They used an 11.1 m-long model of the surface ship and LDV for measuring 2 velocity field components behind the propeller in the presence of the rudder. Propeller-induced velocities were computed using a panel method (Hsin et al., 1991). The velocities in the slipstream were computed using the computer code FPV-10 in conjunction with the panel method. Rudder forces and pressure distributions were computed using the VSAERO code with 50 chordwise panels and 20 spanwise panels. The pressure distributions at = 0 with and without the propeller are shown in Figure 8.6. The authors have found a significant influence of the propeller on the rudder lift and a minor impact on its drag. In practical terms, the presence of the propeller corresponded to rotating the rudder to an angle of attack of about 2.9.
1.2 1.0 0.8 0.6 0.4 0.2 0 0.2
With propeller
Without propeller

Figure 8.5 Cavitation on a spade rudder provoked by a propeller at zero rudder angle, path of tip vortices (courtesy of VWS). Additionally Kracht identified rudder tip cavitation, which could produce erosion of rudder tip, rudder gap cavitation, and cavitation near irregularities on the rudder surface. As a measure against rudder cavitation Kracht also considered endplates, flaps and a leading edge rotating cylinder. Shen et al. (1997a) and Shen et al. (1997b) are two parts of one report describing design, analysis and testing of a twisted rudder to improve rudder cavitation. Ship trial observations and dry-dock inspection data is included. In the first part Shen et al. (1997a) investigate cavitation properties of the rudder behind the propeller. First, with reference to Jiang et al. (1995), they described a case of rudder erosion on a Navy combatant ship. Rudder cavitation was observed at 23 knots. With

Inboard Outboard


0.2 x/c




Figure 8.6 Pressure distribution at = 0 with and without propeller. It was demonstrated that for accurate rudder lift estimations one should take into account the wake profile on the hull. Observations of extensive rudder cavitation in the William B. Morgan Large Cavitation Channel (LCC), shown in Figure 8.4 and full-scale erosion in Figure 8.8, showed that starting at 26 knots, cavitation was unavoid-

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


able. The full-scale prediction of cavitation inception based on model tests in the LCC may be seen in Figure 8.9 as a combined diagram for both rudders showing minimum angles at which cavitation incepts. At 20 knots there was no cavitation inception on the inboard surface up to +18 whereas on the outboard surface up to -8.5 (+ means the trailing edge is deflected outward away from the ship centreline).

Figure 8.8 Full-scale rudder cavitation erosion.




0 20








Figure 8.9 Predicted surface cavitation inception envelope on fleet rudder. Figure 8.7 Rudder cavitation at the William B. Morgan Large Cavitation Channel, LCC (VS = 31 kts, = 0 deg). Shen et al. (1997b) additionally investigated the application of a twisted rudder to enhance cavitation performance. The maximum twist angle was 5.2 at 66% of the span from the root chord. Surface cavitation inception characteristics are compared in Figure 8.9. and show that the twisted rudder is superior to the non-twisted rudder. Table 8.2 Surface cavitation inception on the port rudder.
Measured Rudder Angles (LCC) in deg Ship Cavitation speed Non-Twisted Rudder Twisted Rudder number () (knots) Inboard Outboard Inboard Outboard 20 2.81 >18 -8.5 14.5 -15.1 23 2.12 >14.5 -4.9 11.8 -12.3 25 1.80 13.4 -3.0 9.0 -10.2 31 1.17 (no data) (no data) 2.3 -2.5 Predicted Rudder Angles (Fleet) in deg Ship Cavitation speed Non-Twisted Rudder Twisted Rudder number () (knots) Inboard Outboard Inboard Outboard 20 2.81 >20 -6.5 16.5 -13.1 23 2.12 >16.5 -2.9 14.8 -10.3 25 1.80 15.4 -1.0 11.0 -8.2 31 1.17 (no data) (no data) 4.3 -0.5


The Propulsion Committee

23rd International Towing Tank Conference


so that small effects of hydroelasticity have become embedded within model-full scale correlation. Some work has been advanced in the prediction of hydroelastic effects on propellers in backing conditions and further advances have been made in analysis of composite propellers. No new experimental techniques have been identified for the measurement of blade deflection and strain.
9.2. General Finite Element Analysis (FEA)

Propeller hydroelastic effects were most recently addressed by the 19th ITTC Propulsor Committee in 1990. At that time, with the advent of highly skewed propellers, elastic effects had been identified to affect powering performance and in some cases ultimately lead to blade structural failure primarily in backing. By 1990, in a number of instances, finite element structural analysis had been applied to propellers. Iterative hydrodynamic and structural analysis had been applied by Atkinson & Glover (1988) to show changes in blade pitch under load and resulting increases in rpm on the order of 37%. Atkinson and Glover used lifting surface models to compute blade loads, and relatively few, thick shell elements to represent the blade. In the last ten years, advances have been made in the computation of propeller blade loading using panel methods and RANS computations. Finite element analysis (FEA) has been advanced with the capability to utilize many elements, typically up to 10,000, often as solid quadrilateral elements. With these advances, solving the elastic problem iteratively, would appear straight forward, but only a few researchers have addressed the coupled problem. As was stated in the 1990 report, with the advent of balanced skew, and the adoption of more moderately skewed propellers, the degree of pitch change under load has been less of a problem. The amount of pitch change has been small, relative to the combined uncertainty in the powering prediction,

Although not addressing the hydroelastic problem specifically, a number of researchers have presented advancements and applications of FEA to propeller problems, and are summarized below. Brockett & Brockett (1994) presented a detailed account of application of NASTRAN to a number of experimental data sets, including uniform air pressure loaded laboratory experiments, and the 18th ITTC comparative calculation on a point loaded skewed propeller.
9.3. Propeller Hydroelastic Effects

Jiang et al. (1991) presented a calculation procedure for propellers in backing and crashback conditions. The 22nd ITTC Propulsion Committee reviewed this work in some detail except for interactive computations accounting for blade deflection. This was performed for steady backing conditions utilizing NASTRAN for FEA and lifting surface computations using PSF3 for the blade loading predictions. Blade section drag coefficients were set to 0.02. A comparison was made with data for propeller 4383 operated in the astern condition. Propeller 4383 is a five bladed highly skewed propeller with 72 degrees of skew at the tip with a linear skew distribution and is shown in Figure 9.1.

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


the blade. The velocity field and structural deformation was solved simultaneously, using Bernoullis equation to determine blade pressure. A Newton-Raphson procedure was used to solve the coupled equation, adopting three structural convergence criteria, blade displacement, external forces, and external work, and two criteria on the fluid calculation, the wake velocities at the trailing edge and the singularity strength at each iteration.

Figure 9.1 Propeller 4383 (D = 304.5 mm). The propeller was tested in backing at constant J over a range of tunnel speeds at constant cavitation number with results shown in Figure 9.2. The thrust and torque monotonically increased with increasing dimensional load. The procedure outlined above predicted the rate of increasing thrust, as shown below. The rate of torque increase was underpredicted. The opposite thrust, torque trend was observed when the computations were made with zero blade thickness. Jiang argued that to better simulate viscous flow at the blunt, rounded trailing edge, (the leading and trailing edges are reversed in backing) computing the zero thickness case would better simulate flow separation at the trailing edge. The amount of pitch change for the condition computed was not presented. The uncertainty in prediction of the backing loads including hydroelastic effects is potentially due to combined errors from the load and deflection computations. Jiang showed poor prediction of axial deflection for P4383 under uniform loading, which contributes most to blade pitch change. Lin & Lin (1996) present a coupled nonlinear hydroelastic analysis, which produces converged solutions for very low J conditions. The lifting surface method of Kerwin and Lee (1978) was coupled directly with a nine node shell element FEA representation of

Figure 9.2 Effect of blade deflection on thrust and torque in backing (Jiang et al., 1991). An example was presented for a moderately skewed propeller compared against the method run in a linearized fashion. The linear approach assumes the structure remains rigid when the fluid response is calculated. The resulting pressure is imposed on the structure to obtain the structural response. This process is iterated to convergence. For the linearized case, the nonlinear convergence criteria are met, except for equilibrium between internal and external forces. The example case, run at J=0.1 showed divergent results for the linearized method, and convergent results for the


The Propulsion Committee

23rd International Towing Tank Conference

coupled nonlinear method. Figure 9.3 shows the convergence history and Figure 9.4 shows the pitch deformation for the sample case.

unskewed props, but over predictions of stress occurs on the skewed prop. For the skewed case, strains correlate well in the axial direction, but are under predicted in the rotational direction. The axial strain contributes most to pitch changes, resulting in improved pitch change prediction over Jiang et al. (1991). Hydroelastic interaction is determined by establishing the deformed blade surface, computing a new surface pressure distribution, and recalculating the structural deformation. An integrated load (KT, and KQ) is used as a convergence criterion, usually requiring few iterations. Blade thickness is a design parameter routinely explored in the computation. When computing for thin blades, often resulting in large deformation, the initial and updated pressure distributions are relaxed to stabilize the computation. Two full-scale propellers (Figure 9.5) were analyzed, with comparison to full scale stress measurements performed at 0.7, and 0.9 radius. Measurements compared well to computations at the design advance coefficient.

Figure 9.3 Convergence history for MAU 360 propeller.

Figure 9.4 Resultant MAU 3-60 propeller.

deflected pitch for Figure 9.5 Conventional (CP) and highskew (HSP) full scale propellers. Full scale hydroelastic calculations were performed at design J of 0.62 for CP and 0.66 for HSP propellers and reduced J. Insignificant deflection was seen at design J. At J=0.1, Figure 9.6 shows a significant pitch change for the two propellers. The CP propeller, with little skew, showed increased pitch, while the HSP propeller showed pitch reduction, typical of highly skewed propellers.

Georgiev & Ikehata (1998) presented a procedure to account for hydroelastic effects incorporating a panel method coupled with a thick shell FEA. The panel method incorporates a bicubic polynomial representation of the blade surface. The thick shell FEA incorporates a 16 node curved isoparametric hexahedron. Typical computations utilized 320 elements. Validation of computed stress and strain is shown with uniform air pressure loading data on DTMB propellers 3005, 3604, 4381 (unskewed) and propeller 4383 (skewed, Figure 9.1). Stress predictions correlate well with the

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


urethane leading and trailing edges. The outer 20% of the propeller radius was specified metallic, Nickel-aluminium bronze (NAB). FEA was performed with 20 and 15 node curved solid elements, carefully specifying the appropriate material properties, and interface connectivity. The relative performance is shown in the table below. The composite blade has higher stresses, larger deflections, but half the weight of a solid bronze blade. The large deflections of the composite blade may require consideration to check powering performance, but no hydroelastic computations were made in the study.
Bending Stress Shear Stress Max Deflection Relative weight Solid NAB 1 1 1 1.0 Composite 1.65 1.36 10 0.5

Figure 9.6 Pitch change due to blade deflection at J = 0.1. The resultant load change due to deflection is shown in Figure 9.7 for open water prediction. No loading effect is seen at design J, while small thrust loss is seen with the HSP propeller.

Strmberg (1991) has developed a composite propeller that deforms under load to improve cavitation and noise performance. The composite propeller, Flexprop, has been demonstrated at sea on a 460 bhp RSwN minesweeper and was shown to reduce shaft torque fluctuations. Few details have been provided on the composite design. It is possible that dynamic effects for these applications may be important or dominant in affecting cavitation performance. Searle et al. (1994) presented a description of ongoing work in the development of composite propellers directed towards small propellers. A resin transfer melding process was described, detailing techniques to ensure low void fraction using carefully tooled vacuum melds. Fiber orientation for pitch reduction under load was explained. Model tests of four composite propellers were compared against a bronze geosim and showed that the composite propeller had equivalent open water efficiency near design, and improved efficiency at low J, where blade deflection was assumed to reduce pitch. Model composite props were constructed with bronze hub inserts. Sea trials were con-

Figure 9.7 HSP propeller thrust loss due to blade deflection.

9.4. Composite Propellers

Lin (1991) presented a FEA of a thick shell, foam core composite propeller. The structural performance of the composite propeller was compared to a conventional bronze propeller of the same shape. The composite propeller consisted of a triaxial E-glass shell, a braided E-glass shear web, and nonstructural


The Propulsion Committee

23rd International Towing Tank Conference

ducted on a 0.5 m diameter propeller demonstrating successful in-service performance. Dai et al. (1995) presented an interesting hydrofoil design concept to improve tip cavitation performance for a propeller blade operating in a nonuniform wake field. A composite blade was designed with chordwise and spanwise varying material stiffness properties, such that under load, the blade tip twists to reduce tip loading. An elliptical threedimensional hydrofoil of aspect ratio 2.55 was selected for the design. From the blade loading computed at an 8 degree angle of attack, the composite foil deformation was computed with FEA to arrive at a sufficient deformation to impact cavitation performance. With the specified spanwise blade twist and deflection normal to the foil planform, a detailed panel method computation of the rigid and deformed geometry demonstrated significant reduction in near-tip leading edge suction peaks and field pressures in the tip vortex downstream of the blade. The foil was described as a sandwich construction, with a nonstructural center core and an outer thick shell with varying orientations of S glass/epoxy plies. Unfortunately, no details were provided on the internal structural layout. Gowing et al. (1998) presented design information and model test results for the foil concept above. An elliptical planform foil of 3.5 aspect ratio was tested in the NSWCCD 24 water tunnel for load and cavitation performance. The foil designs were a sandwich construction with the structural carbon fiber cloth oriented 10 and 15 degrees from the spanwise axis of the foils, which produced foil twist under load resulting in decreased tip loading with increased mean loading. A third foil was built of solid aluminum. Load measurements showed reduction in lift slope with increased mean load, most pronounced with the 15 degree laminate foil as shown in Figure 9.8.

Cavitation tests for the aluminum foil, and the 10 and 15 degree laminate foils showed significant improvement in tip vortex performance, shown in Figure 9.9. No deflection measurements were made.

Figure 9.8 Change in lift slope with increasing load.

Figure 9.9 Cavitation inception performance of composite foils. An interesting development is the use of rapid prototyping processes to easily produce plastic propellers for potential use in model testing. The modern resins utilized are reasonably strong, but are subject to water absorption and have limited toughness. With proper sealing of the resin and careful handling, this option for model propellers may become attractive in the future. Model propellers subject to heavy loading may not be suitable, or may require hydroelastic analysis before use.

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


9.5. Recommended Procedures to Account for Hydroelastic Effects.

For some applications propeller hydroelastic effects require consideration. These would be cases of very heavily loaded propellers, critical off design conditions, such as bollard pull or backing, and the application of composite or plastic propellers. It would be recommended, under these situations, to analyze hydroelastic effects on powering performance using modern analysis coupled with applicable hydrodynamic assessment tools, such as panel or lifting surface methods. FEA should be performed with the computed distributed loads applied to the blade. Iterative adjustment of the blade geometry due to the applied load will generally reach convergence within a few iterations. Care may be required for extreme loading cases, where either the updated pressure distribution should be relaxed through the iterations or more direct nonlinear approaches incorporated. For heavily loaded propellers at ahead conditions, a simple check can be used to estimate propeller blade pitch changes under load, or be used as an indication to proceed with more detailed FEA. The change of the pitch angle may be estimated from the following empirical rule, which has been derived at MARIN from results of finite element calculations:
= 0.74 ( S ,max 13) T D 2 / ( Z t 3c ) where


Earlier ITTC Committees, first the Propulsor Committee, later the Propulsion Committee, compiled and expanded a data base of literature on the subjects covered by these respective Committees. Thanks to their efforts this data base grew to almost 8000 entries. A huge expansion took place in the years just before the present Committee. This was achieved by entering older and recent literature referred to in previous ITTC reports, but covering a much wider area than just marine propellers because of the widened scope of the present and former Propulsion Committees. This data base consists of entries of citations only with some key words. The Propulsion Committee of the 23rd ITTC has decided to stop the activity of further expansion of this data base. The reasons for this decision are as follows. There are several more professional data bases available. The accessibility of these data bases has been improved considerably by modern forms of electronic communication. Worth mentioning in this category are MTA (Marine Technical Abstracts of BMT and the IMarE catalogue) and MARNA (Technical University of Delft, The Netherlands). These professional data bases, though overlapping to a large extent, are far more developed than that of the ITTC Propulsion Committee. They include the abstracts, and involve a retrieval service for the full documents. In the area of marine propulsion there are about 80000110000 entries. Moreover, these databases are kept up to date on a continuous basis. Duplicating the efforts of professional organisations who continuously maintain these much larger data bases and who offer retrieval services of the literature is considered an inappropriate use of resources. The professional or-

T D Z t c

pitch angle reduction, degrees propeller thrust, kN propeller diameter, m number of blades blade thickness at 0.6R, mm blade chord at 0.6R, m

s, max maximum skew angle, degrees


The Propulsion Committee

23rd International Towing Tank Conference

ganisations offer the service of sending on request hard copies of papers, a service which would be new and outside of the aims of the ITTC. Expanding and maintaining the data base has not been a task formally assigned to the Propulsion Committee. The Committee has decided to concentrate on the core of its activities rather than pursuing the further expansion of the present literature data base. Nevertheless, the Propulsion Committee is prepared to make the current literature data base available to those member organisations that show an interest in its use. The Propulsion Committee has interpreted this task, with guidance from the Advisory Council, as the compilation of material which is listed in the section on References.

in the propulsion condition from model scale to full scale. Use of CFD methods is becoming helpful to assess these scaling issues. Ensuring sufficiently accurate model propeller geometry for model scale cavitation testing requires geometry inspection procedures. Because of the small size of blade edges, measurements are difficult. A number of efforts can be proposed to address this issue, as follows:

The state-of-the-art of Surface Piercing Propellers (SPP), as well their application, requires the development of standardized model test procedures. This would include standardized parameters, testing guidelines, and tests optimized for both propeller design and propulsion prediction for ships with SPP. Simultaneously, further development of computer simulation of ventilated propellers should be advanced in order to provide a computationally based SPP design procedure. Propeller designs for high-powered singlescrew mega containerships has become very demanding due to their large diameter and loading levels. To attain an optimum balance between design parameters, improvement in the procedures for the prediction of vibration excitation and cavitation erosion is required. Podded propulsion technology has advanced rapidly in the last decade. Current measuring techniques and procedures for predicting full scale performance vary amongst the different towing tanks. A problem to be solved is the extrapolation of the housing drag

Poll ITTC member organizations on their propeller model inspection procedures. Propose standardized measurement procedures. Conduct a comparative exercise circulating a model propeller amongst ITTC members, measuring the propeller using the suggested methods. Prepare an ITTC Recommended procedure for the inspection of model propellers suitable for both powering and cavitation testing.

Propellers operating in crashback represent the most challenging condition to analyze due to the potential for complex blade flow separation, and unsteady nonperiodic inflow. Advances have been made in the application of steady and unsteady RANS to the crashback condition, but accurate computation of extensive unsteady blade flow separation may require the application of large eddy simulation (LES) modelling to capture the instantaneous blade flow. The use of instantaneous global and local PIV flow measurements along with individual blade loads measurements will provide experimental details for validation. Simple techniques using tufts and cavitation can qualitatively visualize the flow. The Committee recommends use of Grigsons line as a physically more accurate reflection of turbulent flat plate friction. The development of an easy to use version of Grigsons direct algorithm is still necessary to improve upon Grigsons approximate curve fits. It is recommended that member organiza-

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


tions conduct independent studies to assess correlation allowances when using Grigsons line with the eventual goal of its adoption as a standard for model/ship correlation. The results of the Gothenburg 2000 Workshop showed significant progress in resistance calculation by using RANS, but further effort is needed. Results show that from the average of all the computations, the calculated k-factors using CF0 from the ITTC 1957 formula are about 1/3 higher at full scale relative to model scale. This trend does not make physical sense and could be partly explained by griddependency of numerical predictions and questionable accuracy of full-scale flow turbulence modelling in RANS codes. The validity of the ITTC 1957 formula as an accurate friction formula can be questioned. If the Grigson turbulent flat plate friction line is used instead of the ITTC 1957 line to obtain the form factors, the full scale values are on average less than 10% higher than model values. The use of a rational friction line brings model to full scale discrepancy closer. It is suggested that computations be made of the CF0 values for a flat plate in order to better understand capability of CFD to compute form factor. Extrapolation of open water and propulsion results from model scale to full scale is not satisfactorily addressed by current procedures. Methods based on turbulent flow stimulation as flow tripping and leading edge roughening look promising but none of them is mature enough to be accepted as a standard in routine experiments. Scaling laws that are used for conventional propellers are not adequate to describe complex propulsors. The nature of the interactions between the propeller itself and the unit passive components is still not fully understood and extensive investigations are still required to develop reliable procedures. State-of-the-art RANS computations represents a powerful tool to investigate Rey-

nolds number effects on propulsive parameters, but further improvements are still required to obtain reliable extrapolation laws from hydrodynamic codes. Lifting surface methods are well established for the evaluation of forces and moments of marine propulsors. Efficiency of lifting surface methods having short computing times, makes the methods popular for iterative evaluation of propeller performance, such as optimization of propeller geometry. The inherent inaccuracy of the methods, for the pressure distribution near blade leading edge and blade tip region, limits the use of the methods for cavitation analysis. Panel methods are widely used for the performance prediction of propellers with an increasing level of confidence, while requiring reasonable computing time. RANS methods are maturing and capable of predicting steady performance and pressure distributions on the blades of marine propellers for design and off-design conditions. Application of the methods includes ducted propellers, podded propulsors and integrated propulsors. Extension of the methods for propellers operating in an unsteady circumferentially varying inflow field requires significantly more computing time, but is beginning to be addressed. The impact of parametric variations of the propeller-rudder system upon propulsive performance can be estimated adequately using simplified computation methods. It has been demonstrated that open-water propellerrudder interaction is very similar to that behind the ship hull. Results of reviewed investigations are suitable for design decisions concerning propeller/rudder respective positions, rudder thickness, etc. Nevertheless, for practical design purposes, it is still reasonable to recommend that propeller-hull interaction effects be estimated from ship model propulsion testing.


The Propulsion Committee

23rd International Towing Tank Conference

The number of publications discussing the cavitation aspects of propeller-hull interaction is limited. Principal reported results indicate the potential of rudder cavitation erosion to occur at speeds over 2023 knots at non-zero steering angles and to be unavoidable at speeds beyond 25 knots even when sailing straight ahead. Some progress has been made in the analysis of hydroelastic effects on propellers. For metallic propellers, hydroelastic effects on performance are minimal at the design operating condition. More significant effects are seen at off design conditions, but without precise off-design requirements, computations of elastic effects may not be useful. Composite propeller technology has advanced in the last ten years. The large deflections inherent with composite blades have been shown to potentially reduce unsteady blade loading. The Committee has decided to discontinue the database developed by the previous Propulsion and Propulsor Committees. Existing professional databases covering the marine field are believed to provide a better service in this regard. The Committee will compile references referred to in the report to meet the requirements of this task.


Abdel-Maksoud, M., and Heinke, H.-J., 1999, Investigation of Viscous Flow around Modern Propulsion Systems, Proc. of the International CFD Conference, CFD99, Ulsteinvik, Norway. Achkinadze, A.S., 1982, Theoretical Investigation of the Conditions of Air Penetration to Lateral Thruster Inlet (in Russian), Proceedings of Leningrad Shipbuilding Institute, Leningrad, Soviet Union, pp. 15-26. Achkinadze, A.S., 1989, Design of optimal screw propellers, turbines and freely rotating turbopropellers adapted for radially nonuniform swirled flow, Proc. Fourth Internat. Symposium on PRADS89, 2328 October 1989, BSHC, Varna, Bulgaria, vol. 3, pp. 124/1-13, 1989. Achkinadze, A.S., 2001, Supercavitating propellers, Preprint: Lecture Series Monographs, Von Karman Institute for Fluid Dynamics, RTO AVT/VKI Special Course: Supercavitating Flows, 12-16 February 2001, Rhode-Saint-Gense, Belgium. Achkinadze, A.S., and Krasilnikov, V.I., 1999, A numerical lifting surface technique for account of radial velocity component in screw propeller design problem, Proceedings of the Seventh International Conference on Numerical Ship Hydrodynamics, 19-22 July 1999 Nantes, France, pp. 3.3-1- 3.3-15. Achkinadze, A.S., and Krasilnikov, V.I., 2001, A New Velocity Based BEM for Analysis of the Non-cavitating and Cavitating Propellers and Foils, Proceedings of International Symposium on Ship Propulsion (SP2001: Lavrentiev Lectures), 19-21 June 2001, St. Petersburg, Russia, paper 4, pp. 45-57.

Adopt the amended Procedure Propulsion, Cavitation, Model-Scale Cavitation Test 7.5-02-03-03.1 Adopt the amended Procedure Propulsion, Cavitation, Description of Cavitation Appearances 7.5-02-03-03.2 Adopt as an Interim Procedure Propulsion, Performance, *Podded Propulsor Tests and Extrapolation 7.5-02-03-01.3

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


Achkinadze, A.S., and Krasilnikov, V.I., 2001a, A new velocity-based BEM for analysis of the non-cavitating and cavitating propellers and foils, SP2001: Lavrentiev Lectures, 19-21 June 2001, St. Petersburg, Russia. Achkinadze, A.S., and Krasilnikov, V.I., 2001b, A velocity-based boundary element method with modified trailing edge for prediction of the partial cavities on the wings and propeller blades, 4th International Symposium on Cavitation, CAV2001, 20-23 June 2001, Pasadena, CA, USA. Achkinadze, A.S., Krasilnikov, V.I., and Stepanov, I.E., 2000, Hydrodynamic design procedure for multi-stage blade-row propulsors using generalized linear model of the vortex wake, SNAME Propellers/Shafting 2000, Virginia Beach, VA, USA. Achkinadze, A.S., Berg, A., Krasilnikov, V.I., and Stepanov, I.E., 2001, Numerical and experimental verification of the SPA/QSPA-POD velocity-based BEM program for steady and quasi-steady analysis of the podded propellers, SP2001: Lavrentiev Lectures, 19-21 June 2001, St. Petersburg, Russia. Ando, J., and Nakatake, K., 2001, Calculation of three-dimensional unsteady sheet cavitation by a simple surface panel method-SQCM, 4th International Symposium on Cavitation, CAV2001, 20-23 June 2001, Pasadena, CA, USA. Ando, J., Kanemaru, T., Ohashi, K., and Nakatake, K., 2000, Calculation of threedimensional unsteady sheet cavitation by a simple surface panel method, J. of SNAJ, Vol. 188. Ando, J., Matsumoto, D., Maita, S., Ohashi, K., and Nakatake, K., 1999, Calculation

of three-dimensional steady cavitation by a simple surface panel method, J. of SNAJ, Vol. 186. Anon, 1981, MARIN Report No. 7, March 1981. Anon, 2000 Five Giant Liners from Hyundai for Costamare, The Naval Architect, June 2000. Arnone, A. Milani, A., Caprino, G., and Traverso, A., 2000, 3D Viscous Calculations of Marine Propeller Flows for Practical Applications, Proc. of the International Conference on Ship and Shipping Research, NAV 2000, pp. 6.1.1-6.1.10. Atkinson, P., and Glover, E.J., 1988, Propeller hydroelastic effects, SNAME Propellers Symposium 1988, Paper 21. Basin, A.M., 1958, The Influence of Propeller Submersion on Hydrodynamic Characteristics, Proceedings of the University of Water Transport, Vol. 6, Leningrad, Soviet Union (in Russian). Basin, A.M., and Goshev, G.A., 1963, Experimental Investigation of SPP Characteristics, Proceedings of the University of Water Transport, Vol. 45, Leningrad, Soviet Union (in Russian). Bazilevski, Y.S., 2001, On the Propeller Blade Turbulization in Model Tests, SP2001: Lavrentiev Lectures, 19-21 June 2001, St. Petersburg, Russia, pp. 201-206. Betz, A., 1938, Zur Theorie der Leitapparate fr Propeller, Ingenieur-Archiv, Vol. IX. Bindel, S., 1996, Comparison between model and ship cavitation an assessment of available data, Proceedings 11th ITTC, Tokyo.


The Propulsion Committee

23rd International Towing Tank Conference

Bolotin, F.F., 1991, Concerning One Universal Law for Surface Piercing Propulsors, (in Russian), Reports of a Scientific and Technical Conference on Ship Theory, KSRI, Part 2, pp. 220-227. Boorsma, A., 2000, Improving full scale ship powering performance predictions by application of propeller leading edge roughness, Master of Science Thesis, University of Technology, Delft, December 2000. Bose, N., 2002, An implementation of Grigsons friction line, Ship and Ocean Technology, SHOT 2002, Indian Institute of Technology, Kharagpur, India. Brandt, H., 1973, Modellversuche mit Schiffspropellern an der Wasseroberflche, Schiff und Hafen, Bd. 25, No. 5, pp. 415-422, No. 6, pp. 493-504, May 1973. Brockett, T., and Brockett, T., 1994, Practical Experience with Finite Element Analysis of Propeller Blades, SNAME Propellers/Shafting 94, Virginia Beach, VA, USA. Caponnetto, M., 2000, Optimisation and Design of Contra-Rotating Propellers, Proc. of Propellers/Shafting 2000 Symposium, pp. 3/1-3/9. Caponnetto, M., and Rolla, P., 1999, Theoretical and Practical Experience in Developing and Tuning Codes for the Design & Analysis of Propellers, Proc. of the International CFD Conference, CFD99, Ulsteinvik, Norway. Chen, H-H., 1967, The Experimental Determination of the Rudder Forces and Rudder Torque of a Mariner Class Ship Model, College of Engineering, University of California, Report No. NA-67-5.

Chen, B., 1996, Computational Fluid Dynamics of Four-Quadrant Marine Propeller Flow, M.Sc. Thesis, The University of Iowa. Chen, B., and Frederick, S., 1999, Computational Fluid Dynamics of Four-Quadrant Marine-Propulsor Flow, J. Ship Research, Vol. 43, No. 4. Chen, B., 2000, RANS Simulations of Tip Vortex Flows for a Finite-Span Hydrofoil and a Marine Propulsor, Ph.D. Thesis, the University of Iowa, Department of Mechanical Engineering. Cho, C.-H., Kim., G.-D., and Lee, C.-S., 1999, Analysis of two-dimensional hydrofoil problems using higher order panel method based on B-splines, J. of SNAK, Vol. 36, No. 4, November 1999. Choi, J.-K., 2000, Vertical inflow-propeller interaction using an unsteady threedimensional Euler solver, Ph.D. Thesis, The University of Texas at Austin. Choi, J.-K., and Kinnas, S., 2001, Prediction of non-axisymmetric effective wake by a three-dimensional Euler solver, J. of Ship Research, Vol. 45, No. 1. Dai, C., Fraser, J., Coffin, P., Garala, H., and Rappl, G., 1995, Hydrodynamic Simulation of a Passive Blade Control for Tip Vortex Cavitation Control, PROPCAV95, International Conference on Propeller Cavitation, Newcastle Upon Tyne, United Kingdom. Dang, J., 2001, Numerical simulation of unsteady partial cavity flows, Ph.D. Thesis, Delft University of Technology. Davoudzadeh, F., Taylor, L.K., Zierke, W.C., Dreyer, J.J., McDonald, H., and Whitfield, D.L., 1997, Coupled Navier-Stokes and Equations of Motion Simulation of Sub-

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


marine Maneuvers, Including Crashback, Proc. ASME Fluids Engineering Division Summer Meeting, FEDSM97-3129, 22-26 June 1997, Vancouver, BC, Canada. Ea, L., and Hoekstra, M., 2000, Numerical Prediction of Scale Effects in Ship Stern Flow with Eddy-Viscosity Turbulence Models, Proc. of the 23rd Symposium on Naval Hydrodynamics, Val de Reuil, France. Egorov, M.V., 1932, Surface Piercing Propeller Tests in Towing Tank, Proceedings of Institute of Water Transport, Vol. 1, Nikolaev, 1932 (in Russian). Esposito, P., Salvatore, F., Di Felice, F., and Ingenito, G., 2000, Experimental and numerical investigation of the unsteady flow around a propeller, 23rd Symposium of Naval Hydrodynamics, Val de Reuil, France. Esposito, P.G., and Salvatore, F., 2000, Optimal Design of Cavitating Blade Sections by Genetic Algorithms, Proc. of NCT50, International Conference on Propeller Cavitation, pp. 275-288. Ferrando, M., 1997, Surface Piercing Propellers: State of the Art, Oceanic Engineering International, Vol. 1, No. 2, pp. 40-49. Ferrando, M., Panarello, A., Scamardella, A., and Viviani, M., 2000, Surface Piercing Propellers for Displacing Vessels, Proc. of the International Conference on Ship and Shipping Research, NAV 2000, pp. 6.2.1-6.1.16. Ferrando, M., Scamardella, A., Bose, N., Liu, P., and Veitch, B., 2001, Performance of a family of Surface Piercing Propellers, Paper No. W283, Transaction of RINA.

Friesch, J., 2001, Investigations of Podded Drives in a Large Cavitation Tunnel, PRADS 2001, Shanghai, China. Georgiev, D.J., and Ikehata, M., 1998, Hydroelastic Effects on Propeller Blades in Steady Flow, J. SNAJ, Vol. 184. Ghassemi, H., and Allievi, A., 1999, A Computational Method for the Analysis of Fluid Flow and Hydrodynamic Performance of Conventional and Podded Propulsion System, Oceanic Engineering International, Vol. 3, No. 1, pp. 101-105. Gowing, S., Coffin, P., and Dai, C., 1998, Hydrofoil Cavitation Improvements with Elastically Coupled Composite Material, 25th American Towing Tank Conference, Iowa City, IA, USA. Grigson, C.W.B., 1993, An Accurate Smooth Friction Line for use in Performance Prediction, Transactions of the RINA. Grigson, C.W.B., 1999, A Planar Friction Algorithm and its Use in Analysing Hull Resistance, Transactions of the RINA. Gutsche, F., 1967, Einfluss der Tauchung auf Schub und Wirkungsgrad von Schiffspropellern, Schiffbauforschung, 6, No. 5/6, pp. 256-277. Guner, M., and Atlar, M., 1999, A rational approach to the design of Propeller/Stator Combination, International Shipbuilding Progress, Vol. 46, No. 447. Hadler, J.B., and Hecker, R., 1968, Performance of Partially Submerged Propellers, 7th Symposium on Naval Hydrodynamics, August 1968, Rome, Italy, S.25-30, pp. 1889-1484. Hashimoto, J., Nishikawa, E., and Li, Z., 1983, An Experimental Study of Propel-


The Propulsion Committee

23rd International Towing Tank Conference

ler Air Ventilation and its Induced Vibratory Forces, Bulletin of the Marine Engineering Society in Japan Quarterly, Vol. 11, No. 1, pp. 11-21. Heinke, M.J., 2001, Alternative Propulsion Concepts for Fast Navy Ships, Part II: Podded Drives for Navy Ships, STG Sprechtag Unkonventionnelle Rumpfformen fr Marineschiffe, September 2001, Potsdam, Germany. Holtrop, J., 2001, Extrapolation of Propulsion Tests for Ships with Appendages and Complex Propulsors, Marine Technology, Vol. 38, pp. 145-157. Hsin, C.-Y., Kerwin, J.E., and Kinnas, S.A., 1991, A panel method for the analysis of the flow around highly skewed propellers, SNAME Propellers/Shafting 91, Virginia Beach, VA, USA, pp. 11.111.13. Hughes, G., 1963, The influence of form and scale on model and ship resistance, 10th ITTC Proceedings, Teddington, United Kingdom, Vol. II. Iljin, V.M., 1975, The Investigation of Screw Hydrodynamic Characteristics at Partial or Lightly Submerged Conditions, (in Russian), KSRI Proceedings, Vol. 285, pp. 143-154. Isay, W.H., 1965, ber die Wechselwirkung zwischen Schiffsruder und Schraubenpropeller, Schiffstechnik, Vol. 12, No. 62. Ishida, S., 1987, The Recovery of Rotational Energy in Propeller Slipstream by Fins Installed after Propellers, Naval Architecture and Ocean Engineering, Vol. 25. Jang, T.S., Kinoshita, T., and Hino, T., 2001, An Optimization Method Based on Hilbert Space Theory for Design of Marine Propellers and Hull Forms, Proc. of the

8th Int. Symposium on Practical Design of Ships and Other Floating Structures, pp. 699-704. Jiang, C.W., Huang, T.T., Ng, R., and Shin, Y.S., 1991, Propeller Hydrodynamic Loads and Blade Stresses and Deflections During Backing and Crashback Operations, SNAME Propellers/Shafting 91, Virginia Beach, VA, USA. Jiang, C.W., Remmers, K.D., and Shen, Y.T., 1995, Rudder Cavitation Studies at DTMB Large Cavitation Channel, International Symposium on Cavitation, Deauville, France. Jiang, C.W., Dong, R.R., Liu, H.L., and Chang, M.S., 1996, 24-inch Water Tunnel flow Field Measurements During Propeller Crashback, 21st ONR Symposium on Naval Hydrodynamics, Trondheim, Norway, pp. 136-146. Karafiath, G., 1998, Hydrodynamic Performance with Pod Propulsion US Navy Experience, 25th ATTC, September 1998, Iowa City, IA, USA. Karim, M.M., and Ikehata, M., 2000, A Genetic Algorithm (GA) Based Optimization Technique for the Design of Marine Propellers, Proc. of Propellers/Shafting 2000 Symposium, pp. 16/1-16/12, 2000. Kawakita, C., and Hoshino, T., 1998, Design System of Marine Propellers with New Blade Sections, 22nd Symposium on Naval Hydrodynamics, Washington, DC, USA. Kempf, G., 1934, Immersion of Propellers, Transactions of the North East Coast Institution of Engineers & Shipbuilders, NECIES, Vol. 50, pp. 6, April 1934. Kerwin, J.E., and Lee, C.S., 1978, Prediction of steady and unsteady marine propeller

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


performance by numerical lifting surface theory, SNAME Transactions, Vol. 86, pp. 218-253. Kinnas, S.A., 2001, Supercavitating propellers, Preprint: Lecture Series Monographs, Von Karman Institute for Fluid Dynamics, RTO AVT/VKI Special Course: Supercavitating Flows, 12-16 February 2001, Rhode-Saint-Gense, Belgium. Korshunov, A.A., and Eroshevskyi, J.A., 1984, The Calculation of the Beginning of SPP Aeration (in Russian), Krylov Science-Technical Society, Vol. 400, pp. 22-30. Kracht, A., 1995, Cavitation on Rudders, PropCav 95, May 1995, Newcastle upon Tyne, United Kingdom. Kracht, A., 1997, Propeller und Ruder als Propulsionsorgan, Schiffbautechnische Gesellschaft, Vol. 91. Kudo, T., Ukon, Y., and Kato, H., 1999, Study on theoretical design method of trans-cavitating propeller, J. of SNAJ, Vol. 186. Larsson, L, Stern, F., and Bertram, V. (Eds.), 2000, Proceedings of Gothenburg 2000, Workshop on Numerical Ship Hydrodynamics, Gteborg, Sweden, 2000. Lee, H., and Kinnas, S.A., 2001, Modelling of unsteady blade sheet and developed tip vortex cavitation, 4th International Symposium on Cavitation, CAV2001, 20-23 June 2001, Pasadena, CA, USA. Lee, C.-S., and Kerwin, J.E., 2001, A Bspline based high order panel method, J. of Ship Research, Accepted for publication.

Lehman, A.F., and Romandetto, R., 1968, An Experimental Study of the Unsteady Forces Induced on a Rudder due to a Propeller Acting in a Wake, Oceanics, Inc. Report No. 68. Lewis, F.M., 1969, Propeller Vibration Forces in Single Screw Ships, Trans. SNAME, Vol. 77. Li, D.-Q., 1993, A Linear Method for the Calculation of Propeller-Rudder Interaction, Chalmers University of Technology, Report No. 28, Gothenburg, Sweden. Li, D.-Q., 1993, Study on the PropellerRudder Interaction Based on a Linear Method, Chalmers University of Technology, Report No. 31, Gothenburg, Sweden. Li, D.-Q., 1994, Investigation on the Propeller-Rudder Interaction by Numerical Methods, Doctor Thesis, Chalmers University of Technology, Gothenburg, Sweden. Li, D.-Q., and Dyne, G., 1995, Study of Propeller-Rudder Interaction Based on a Linear Method, Int. Shipbuild. Progr., Vol. 42, No. 431. Lin, G.F., 1991, Three Dimensional Analysis of a Fiber Reinforced Composite Thruster Blade, SNAME Propellers/Shafting 91, Virginia Beach, VA, USA. Lin, H.-J., and Lin, J.-J., 1996, Nonlinear Hydroelastic behavior of Propellers Using a Finite Element Method and Lifting surface Theory, Journal of Marine Science and Technology, Vol. 1, pp..114-124. Lindgren, H., 1955, The Influence of Propeller Clearance and Rudder upon the Propulsive Characteristics, Publ. of Swedish State Shipbuilding Experimental Tank, No. 33.


The Propulsion Committee

23rd International Towing Tank Conference

Lipis, V.B., 1972, Concerning Dynamic Scale Effect of Propeller Atmospheric Cavitation (in Russian), Proceedings of CNIMF, Leningrad, Soviet Union, No. 153, pp. 24-31. Liu, P., and Colbourne, B., 2001, A study of wake discretization in relation to the performance of a propeller panel method, SP2001: Lavrentiev Lectures, 19-21 June 2001, St. Petersburg, Russia. Lobachev, M.P., and Tchitcherine, I.A., 2001, The Full-Scale Estimation for Podded Propulsion System by RANS Method, Lavrentiev Lectures, St. Petersburg, Russia, pp. 39-44. Lobachev, M.P., and Tchitcherine, I.A., 2001, The Full-Scale Resistance Estimation for Podded Propulsion System by RANS Method, Proc. of Int. Symp. on Ship Propulsion, St. Petersburg, Russia. Lotveit, M., 1959, A Study of Rudder Action with Special Reference to Single-Screw Ships, North East Coast Institute of Engineers and Shipbuilders, Vol. 73. Mewis, F., 2001, The Efficiency of Pod Propulsion, Proceedings of HADMAR-2001 Conference, October 2001, Varna, Bulgaria, Vol. 3. Mewis, F., 2001a, Pod drives pros and cons, HANSA, 138. Jahrgang, 2001, Nr. 8, S. 25-30. Michael, T., and Jessup, S., 2001, Effect of Model Propeller Trailing Edge Details on Powering Performance, 23rd ATTC, Webb Institute, Glen Cove, NY, USA. Moon, I.-S., 2000, Performance prediction for waterjet propulsors by surface panel method applying blade tip gap flow model, Ph.D Thesis, Dept. of Naval Ar-

chitects and Ocean Engineering, Chungnam National University, Taejon, Korea. Moore, C., Veitch, B., Bose, N., Jones, S., and Carlton, J., 2002, Multi-component blade load measurements on a propeller in ice, SNAME Annual Meeting, Boston, MA, USA. Moriyama, F., 1981, On the Effect of a Rudder on Propulsive Performance, Journal of the Society of Naval Architects of Japan, Vol. 150. Moriyama, F., and Yamazaki, R., 1981, On the Effect of Propeller on Rudder, Transactions of the West-Japan Society of Naval Architects, No. 61 (in Japanese). Moriyama, F., and Yamazaki, R., 1981a, On the Forces Acting on Rudder in Propeller Slip Stream, Transactions of the WestJapan Society of Naval Architects, No. 62 (in Japanese). Nagau, T., and Tanaka, H., 1974, Model Tests of the Partially Submerged Propeller, Transactions of the West-Japan Society of Naval Architects, No. 47, February 1974. Nakatake, K., Arimura, F., and Yamazaki, R., 1978, Effect a Rudder on Propulsive Performance of a Ship, Transactions of the West-Japan Society of Naval Architects, No. 55 (in Japanese). Nakatake, K., Koga, T.F., and Yamazaki, R., 1981, On the Interaction between Screw Propeller and Rudder, Transactions of the West-Japan Society of Naval Architects, No. 61 (in Japanese). Paterson, E.G., Kim, J., and Stern, F., 1999, Unsteady RANS simulation of an integrated marine propulsor, 17th Conference on Numerical Ship Hydrodynamics, Nantes, France.

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


Pedersen, P.S., 1999, The Future of the LowSpeed, Two-Stroke Engine, Propulsion 2000, The Institute of Marine Engineers, November 1999, London, United Kingdom. Praefke, E., Richards, J., and Engelskirchen, J., 2001, Counter Rotating Propellers without Complex Shafting for a Fast Monohull Ferry, FAST 2001, September 2001, Southampton, United Kingdom. Prischemichin, Y.N., and Korshunov, A.A., 1969, On Model Simulation of Partially or Lightly Submerged Propellers at Bollard Pull Mode, Proceedings of Krylov Science-Technical Society, No. 136, pp. 30-34, Leningrad (in Russian). Pyo, S.-W., and Suh, J.-C., 2000, Modified split panel method applied to the analysis of cavitating propellers, J. of Ship & Ocean Technology, Vol. 4, No. 2. Rose, J.C., and Kruppa, C.F.L., 1991, Methodical Series Model Test Results, Proceedings from the First International Conference on Fast Sea Transportation, Vol. 2, pp. 1123-1148, Trondheim, Norway. Salvatore, F., and Esposito, P.G., 2001, An improved boundary element analysis of cavitating three-dimensional hydrofoils, 4th International Symposium on Cavitation CAV2001, 20-23 June 2001, Pasadena, CA, USA. Sanchez-Caja, A., 1998, DTRC Propeller 4119 Calculations at VTT, 22nd ITTC Propulsion Committee Propeller RANS/Panel Method Workshop, Grenoble, France, 1998. Sanchez-Caja, A., Rautaheimo, P., Salminen, E., Siikonen, T., 1999, Computation of the Incompressible Viscous Flow around a Tractor Thruster Using a Sliding-Mesh Technique, Proc. of The Seventh Interna-

tional Conference on Numerical Ship Hydrodynamics, 1999. Sanchez-Caja, A., Rautaheimo, P., and Siikonen, T., 2000, Simulation of Incompressible Viscous Flow around a Ducted Propeller Using a RANS Equation Solver, Proc. of the 23rd Symposium on Naval Hydrodynamics, Val de Reuil, France, 2000. Sanchez-Caja, A., Martio, J., and Pylkkanen, J.V., 2001, Marine Applications of RANS Solver FINFLO to Propeller and Free-Surface Hull Flows, Proc. of Int. Symp. on Ship Propulsion, St. Petersburg, Russia, 2001. Saunders, H.E., 1957, Hydrodynamics in Ship Design, SNAME, New York, NY, USA. Searle, T., Chudley, J., Short, D., and Hodge, C., 1994, The Composite Advantage, SNAME Propellers/Shafting 94, Virginia Beach, VA, USA. Shen Y.T., Young T., Remmers, K.D., and Chen W., Jiang, 1997a, Effects of Ship Hull and Propeller on Rudder Cavitation, Journal of Ship Research, Vol. 41, No. 3. Shen, Y.T., Jiang C.W., and Remmers, K.D., 1997b, A Twisted Rudder for Reduced Cavitation, Journal of Ship Research, Vol. 41, No. 4., pp. 260-272. Sding, H., 1998, Limits of Potential Theory in Rudder Flow Predictions, Ship Technology Research, Vol. 45. Stanier, M., 1998, The Application of RANS Code to Investigate Propeller Scale Effects, Proc. of 22nd ONR Symposium, Washington, DC, USA, pp. 222-238. Stanier, M., 1998a, Investigation into Propeller skew using a RANS Code. Part 2:


The Propulsion Committee

23rd International Towing Tank Conference

Scale Effects, Int. Shipbuilding Progress, Vol. 45, No. 443, pp. 253-265. Stierman, E. J., 1989, The Influence of the Rudder on Propulsive Performance of Ships, International Shipbuilding Progress, Vol. 36, No. 407, 20th ITTC Proceedings, 1993, Vol. II, The Powering Performance Committee. Strmberg, K.O., 1991, DeformationControlled Composite CP Propeller Blades, Maritime Defence, February 1991. Suhrbier, K.R., 1974, An experimental investigation on the propulsive effect of a rudder in the propeller slipstream. Sugai, K., 1964, On Vibratory Forces Induced on the Rudder behind a Propeller, Ship Research Institute, Tokyo. Suzuki, H., Toda, Y., and Suzuki, T., 1993, Computation of Viscose Flow around the Rudder behind a Propeller. Laminar Flow around a Flat Plate Rudder in Propeller Slipstream, 6th International Conference on Numerical Ship Hydrodynamics, Iowa City, IA, USA. Szantyr, J.A., 2001, Hydrodynamic Model Experiments with Pod Propulsors, Lavrentiev Lectures, St. Petersburg, 2001. Tagory, T., 1963, A Study of the Turbulence Stimulation Device in the Model Experiment on Ship Form, Proceedings of the 10th ITTC, 1963, Vol. II. Taljanova, V., and Stoomph, V., 1976, Investigation of the Impact of Rudder Dimensions and Position on Propulsion of Big Seagoing Merchant Ships, KSRI Report 17159 (in Russian). Tamashima, M., Yang, C.-J., and Yamazaki, R., 1992, A Study of the Flow around a

Rudder with Rudder Angle behind Propeller, Transactions of the West-Japan Society of Naval Architects, No. 83 (in Japanese). Tamashima, M., Yang, C.-J., and Yamazaki, R., 1992, A Study of the Forces Acting on Rudder with Angle behind Propeller, Transactions of the West-Japan Society of Naval Architects, No. 84 (in Japanese). Tamashima, M., Matsui, K., Mori, K., and Yamazaki, R., 1993, The method for the predicting the performance of propellerrudder system with rudder angle and its application to the rudder design, Transactions of the West-Japan Society of Naval Architects, No. 86 (in Japanese). Toyoda, M., Kato, H., and Yamaguchi, H., 1999, A series design of super-cavitating foil sections, J. of SNAJ, Vol. 186. Tsakonas, S., Jacobs, W.R., and Ali, M. R., 1968, A Theory for the Propeller-Rudder Interaction, Stevens Institute of Technology, Report SIT-DL-68-1284; published as Application of the Unsteady-LiftingSurface Theory to the Study of PropellerRudder Interaction, 1970, J. Ship Research, Vol. 14, No. 3. Tsakonas, S., Jacobs, W.R., and Ali, M. R., 1975, Propeller-Rudder Interaction Due to Loading and Thickness Effects, J. Ship Research, Vol. 19, No. 2. Turnock, S.R., 1993, Prediction of Ship Rudder-Propeller Interaction Using a Panel Method, 19th WEGEMT School of Numerical Simulation of Hydrodynamics: Ships and Offshore Structures, Nantes, France. Turnock, S.R., Wellicome, J.F., Molland, A.F., 1994, Interaction Velocity Field Method for Predicting Ship-RudderPropeller Interaction, Propeller & Shaft-

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume I


ing 94 Symposium, September 20-21, Virginia Beach, VA, USA. Tulin, M.P., 2001, The history and principles of operation of supercavitating propellers, Preprint: Lecture Series Monographs, Von Karman Institute for Fluid Dynamics, RTO AVT/VKI Special Course: Supercavitating Flows, 12-16 February 2001, Rhode-Saint-Gense, Belgium, 2001. Vartdal, L., Gjerde, K. M., Bloch, F., and Sittanggang, P. L., 1999, Application of Various CFD methods in the development of the AZIPULL podded propulsion system, Proc. of the International CFD Conference CFD99, 1999. Vasiljev, N., 1996, Investigation of Propeller-Rudder-Hull Interaction, Proceeding of Navy Nowaday Symposium, St. Petersburg, Russia, Section A, Vol. 2. Veikonheimo, T., 2001, The most efficient propulsion system for ULCV, Schiffbauforschung 4/2001, 40. Jahrgang, Rostock, Germany. Voigt, H., 1934, Systematische Schraubenuntersuchungen am Schiffsmodell, Schiffbau. Wang, D., Chai, S., and Wang, Y., 2000, Design Optimization of New Blade Section Based on Genetic Algorithms, Proc. of NCT50 International Conference on Propeller Cavitation, pp. 289-298.

Whicker, L.F., and Fehlner, L.F., 1958, Freestream Characteristics of a Family of Lowaspect-ratio, All-movable Control Surfaces for Application to Ship Design, David Taylor Model Basin, Report No. 933. Yamazake, R., 1968, On the Propulsion Theory of Ships on Still WaterIntroduction, Memoirs of the Faculty of Engineering, Kyushu University, Vol. 27, No. 4, Japan. Yamazake, R., Nakatake, K., and Moriyama, F., 1985, On the Mutual Interaction between the Screw Propeller and the Rudder, reprinted from The Memoirs of the Faculty of Engineering Kyushu University, Vol. 45, No. 1, Fukuoka, Japan. Young, Y. L., and Kinnas, S. A., 2000, Prediction of Unsteady Performance of Surface-Piercing Propeller, Proc. of Propellers/Shafting 2000 Symposium, pp. 7.17.9. Zhu, Z.-L., and Dong, S.-T., 1986, A Theoretical Method for Predicting the Hydrodynamic Performances of PropellerRudder Interaction, International Symposium on Propeller and Cavitation, Wuxi, China. Zondervan, G.-J., and Holtrop, J., 2000, Application of Advanced Sectional Profiles in the Design of Propulsors and Ship Appendages, Proc. of Propellers/Shafting 2000 Symposium, pp. 1/1-1/10, 2000.

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume III


The Propulsion Committee

Committee Chair: Dr. Stewart D. Jessup (NSWC-CD) Session Chair: Dr. William B. Morgan (NSWC-CD)

I. DISCUSSIONS I.1. Discussion on the Report of the 23rd ITTC Propulsion Committee: Form factors for High Speed Craft By: A.F. Molland, University of Southampton, United Kingdom I should like to ask a question about form factors for high speed craft, section 5.4. The propulsion committee points out that, by convention, a form factor (1+k) of unity is used for high speed craft. It goes on to mention some test techniques for medium speed ships. A recommendation for the use of (1+k) = 1.0 for HSMVs is also made by the Specialist Committee on Procedures for Resistance, Propulsion and POW tests, section 3.6. This recommendation tends to arise from factors such as the influence of transom sterns and changes in wetted area on the estimation of form effects for such vessels. It is now well understood that high speed vessels, with relatively fine hull forms, can exhibit significant form effects. The 22nd ITTC Specialist Committee on Model Tests of HSMVs reported alternative methods of obtaining form factors for such vessels, including the use of bow-down tests or by deducting the measured wave pattern resistance from the total resistance at higher speeds. It is apparent that methods do exist for obtaining form factors for HSMVs. Is it not time for ITTC to address this

problem formally? This would entail the investigation and formulation of recommended procedures for the derivation of reliable form factors for HSMVs which are more realistic than (1+k) = 1.0 and which would lead to an extrapolation procedure which is physically more correct.

I.2. Discussion on the Report of the 23rd ITTC Propulsion Committee: Determination of Wake Fraction for Full-Scale Hull with Twin-Skeg Stern from Model Measurement By: Shaoxin Wang, Dalian University of Technology, China Based on the 1978 ITTC method, the wake fraction for full-scale hull can be calculated according to the models result from selfpropulsive test with equi-thrust approach:
Ws = (tm + 0.04) + (Wm (tm + 0.04)) (1+ k)Cfs + Cf (1+ k)Cfm

t s = t m , rs = rm In fact this method can be applied to the hull with single stern only. For use of some hulls with twin-skeg stern it has been found that the difference between wake fraction and thrust deduction from model test is less than 0.04 and the wake fraction for full-scale hull is larger than model one by using the above method. Obviously, this


The Propulsion Committee

23rd International Towing Tank Conference

method cannot apply to the analysis for hull with twin-skeg stern directly. We think that the wake fraction is lower and the thrust deduction is higher as the water velocity and the hull resistance is higher in the area between twin-skegs as compared with a hull with single stern. In the experimental practice the following approach is used at our basin: the propulsive coefficient can be written as that

single roughness parameter to account for the effect of roughness on the correlation allowance for a moderately rough ship hull. Various experiments had shown that a single height parameter was sufficient since moderately rough ship hulls differ little in texture (Townsin, 1990; Townsin & Dey, 1990). For the last 15 to 20 years, Tributyl-Tin Self-Polishing Co-Polymers (TAT-SPC), which can keep a ship free of fouling for 5 years by means of a steady release of the TBT toxin, have dominated the antifoulings market. However, due to environmental side-effects related with TBT, the International Maritime Organisation (IMO) has decided in October 2001 to prohibit the application of TBT-SPCs from 2003 and hence completely phase out their use by 2008. There are currently two alternatives on the market that can also offer 5 years of satisfactory antifouling performance. The first alternative, Tin-free SPC, operate by the same chemical principle but, instead of TBT, gradually leach copper-based toxins that are complemented by booster biocides. The second alternative, Foul(ing) Release coatings, act as a physical rather than a chemical defence against fouling. These coatings are silicone elastomers which have entirely different surface characteristics, notably their surface energy, so that firm attachment of fouling organisms is avoided and the release of the fouling organisms occurs at sufficiently high service speeds (> 15 knots). This contribution summarises the findings of a research project carried out at the University of Newcastle-upon-Tyne to systematically compare the drag, boundary-layer and roughness characteristics of a Foul Release system and a Tin-free SPC system (Candries, 2001) and recommends the ITTC to reconsider the procedure adopted to correlate between roughness and drag. Towing tank experiments have been carried out with two friction Planes of different size. Three series of measurements were carried out for each plane, uncoated, coated with Foul Release

P = 0 h r
in which 0 is the efficiency of open water of propeller; h is the hull efficiency, and

h =

1 ts 1 Ws '
(1 + k )C fs + C f (1 + k )C fm

Ws ' = (t m + q ) + [Wm (t m + q )]

q = 0.01 ~ 0.04, and h is no larger than single stern hull under the same displacement with twin-skeg hulls; r is the relative revolution efficiency; k is the form factor. This correlative calculation should be validated of course. We hope to make a suggestion that ITTC and corresponding technical committees, such as Resistance, Propulsion, or Performance committee, should promote the investigation of the correlative calculation for model-hull with twin-skeg stern from model measurements to ship in some of propulsive factories. I.3. Discussion on the Report of the 23rd ITTC Propulsion Committee: Reconsideration of the correlation of roughness and drag characteristics of surfaces coated with antifoulings By: M. Candries, M. Atlar, University of Newcastle-upon-Tyne, United Kingdom This topic was last considered extensively by the Committee in 1990 (ITTC, 1990a, 1990b). The recommendations of the Powering Performance Committee were to include only a

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume III


and coated with Tin-free SPC. It was found that the Foul Release system exhibits less drag than the Tin-free SPC system. The difference in frictional resistance varied between 2% and 23%, depending on the quality of application (Candries, 2001; Candries et al., 2001). Rotor experiments

were also carried out to measure the difference in torque between coated and uncoated cylinders. The measurements indicated an average 3.6% difference in local frictional resistance coefficient between the Foul Release and Tin-free SPC (Candries et al., 2002a).

Table I.3.1 Overview of the drag characteristics of Foul Release and Tin-free SPC.
Towing tank experiments
2.55 m long plate Sprayed Foul Release Sprayed SPC 6.3 m long plate Sprayed Foul Release Sprayed SPC

(compared to reference, %)
2.0x10 < Re < 4.2x10 3.9 23.4 2.0x107 < Re < 4.0x107 3.9 23.4
6 6


(on average)


Average Roughness (m)

0.20 2.17

44 75

0.21 0.62

62 39

Rotor experiments
Cylinder Sprayed Foul Release Rollered Foul Release Sprayed SPC

(compared to reference, %)
1.0x10 < Re < 2.1x10 4.3 5.7 8.0
6 6


(on average)


Average Roughness (m)

1.00 1.31 1.80

108 218 54

Water tunnel experiments

1m long vertical plate (Emerson Cavitation Tunnel) Sprayed Foul Release Rollered Foul Release Sprayed SPC 1m long vertical plate (CEHIPAR Cavitation Tunnel) Sprayed Foul Release Sprayed SPC

(compared to reference, %)
8.5x103 < Rel < 3.4x104 10.9 13.1 16 1.6x104 < Rel < 4.6x104 14.6 22.9


(on average)


Average Roughness (m)

1.25 1.54 1.80

51 60 69

1.68 2.71

50 30

The friction of a surface in fluid flow is caused by the viscous effects and turbulence production in the boundary layer close to the surface. A study of the boundary-layer characteristics of the coatings was therefore carried out in two different water tunnels using Laser Doppler Ve-

locimetry (LDV). The coatings were applied on 1 m long test sections that were fitted in a 2.l m long flat plate set-up. An outer-layer wall similarity method and the Reynolds stress method were used to determine the friction velocity and both methods showed good agreement with each


The Propulsion Committee

23rd International Towing Tank Conference

other. The experiments indicated that the friction velocity for the Foul Release surfaces is significantly lower than for Tin-free SPC surfaces. This implies that at the same streamwise Reynolds number the ratio of the inner layer to the outer layer is smaller for Foul Release surfaces. The inner layer is that part of the boundary layer where major turbulence production (and hence drag generation) occurs. The roughness functions of the different surfaces, determined from the measured velocity profiles as illustrated in

Figure I.3.1, indicated that on average the Foul Release surfaces exhibit less drag than Tin-free SPC surfaces, which is in agreement with the findings from the towing tank and rotor experiments, as shown in Table I.3.1. No significant differences between the different coatings were found in the turbulence intensities, although this may have been obscured by the experimental precision uncertainties.

Figure I.3.1 Boundary-layer velocity profiles in inner co-ordinates (i.e. the distance from the surface, y+, and the streamwise velocity component U have been scaled by the viscous length scale /U and the friction velocity U respectively) at a free-stream velocity Ue = 5 m/s and at a streamwise location x = 1.607 m from the leading edge. A rollered and a sprayed Foul Release surface were tested to investigate the effect of application method. A surface covered with sand grit was tested in order to have a very rough comparison. The velocity loss or roughness function U+ indicates the difference in frictional resistance between a rough and a smooth surface. (Experimental precision uncertainty over the log-law region: U+: 1.72% for the uncoated steel surface, 1.94% for the rough surfaces; U+: 14.74%). Roughness measurements were carried out on the tested surfaces using a BMT Hull Roughness Analyser. This stylus instrument measured the extreme roughness amplitude over a 50 mm cut-off length at a sampling interval of 1.25 mm, Rt50. For a Foul Release surface, the average of this roughness parameter will not correlate with the measured drag. One of the towing tank experiments and the rotor experiments, for example, indicated that the average roughness was

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume III


higher for the Foul Release surface than for the Tin-free SPC, whereas the measured drag was lower (cfr. Table I.3.1). A detailed roughness analysis of sample plates, coated alongside the tested surfaces and representative of their surface characteristics, was carried out with an optical measurement system fitted with a 3 mW laser. The methodology which was developed to acquire the roughness parameters from six profiles of each sample, applies a moving average boxcar method to filter the data. The upper bandwidth limit or cut-off length was set at 2.5 and 5 mm, the sampling interval at 50 m.

surfaces. The main difference between the Foul Release and the Tin-free SPC systems lies in the characteristics. Whereas the Tin-free SPC surface displays a typical closed texture, the Foul Release surface exhibits a wavy, open texture. This is particularly evident from parameters such as the mean absolute slope a and the Fractal Dimension FD. The spectra of the coated surfaces seem to follow a power law which is dependent on the Fractal Dimension and an implication of self-affine behaviour. A surface with an open texture will have a lower Fractal Dimension than a surface with a closed texture (Candries, 2001). There is relatively little data available in literature on the influence of texture of irregular surfaces on drag, but Grigson (1982) shows that open textures have a beneficial effect on drag. It is thought that the rheology of the paint (which is significantly different for Foul Release systems than for Tin-free SPC systems as is clear from a parameter such as the viscosity) has a direct effect on its texture, whereas amplitudes depend significantly on the application quality. Correlation of the texture parameters with the amplitude parameters, however, shows that the two are inter-related so that bad application can be expected to have a knock-on effect on the texture parameters. The roughness characteristics of both Tinfree and Foul Release surfaces correlate qualitatively with the drag differences given in Table I.3.1 when a texture parameter is included in the roughness characterisation. A semiempirical approach was applied to correlate the roughness characteristics with the drag measurements of the surfaces tested here along with the surfaces included by Townsin & Dey (1990). The approach involved the selection of a characteristic roughness measure h which gives the best correlation assuming that the roughness function of the surfaces follows the Colebrook-White law. The characteristic measure which was found to give the best correlation for the present surfaces is h = Raxa/2 for an effective cut-off length, whereby Ra is the average amplitude (which

Figure I.3.2 Typical roughness measurement of a Foul Release surface.

Figure I.3.3 Typical roughness measurement of a Tin-free SPC surface. Figure I.3.2 and Figure I.3.3 show two typical measurements of a Foul Release and Tin-free SPC surface respectively. The detailed roughness analysis revealed that when the profiles are filtered, the amplitude parameters of the Foul Release surfaces are mostly but not always lower than those of the SPC


The Propulsion Committee

23rd International Towing Tank Conference

strongly correlates with Rt). The effective cutoff length increases with the degree of roughness and varied in the analysis between 2.5 mm for the Foul Release surfaces and 50 mm for a sand grit surface covered with paint (Candries, 2001). The procedure presently adopted by the MC uses the formula suggested by Townsin et al. (1984) to predict the added resistance of new ships from roughness measurements: 103xCF = 44[(h/L)1/3 - 10xRe-1/3] + 0.125 where h is the Average Hull Roughness measured by the BMT Hull Roughness Analyser. Townsin & Dey (1990) showed that the roughness function U/U correlates well with Rt50 for new, moderately rough surfaces (Rt50 < 225 m) and that Rt50 was therefore adequate for quality control as well as for measuring the approximate power penalties due to roughness. Townsin and Dey argued that the reason why a single roughness parameter Rt50, could well predict the added resistance of a wide range of new ship surfaces, is that their texture is fairly similar, allowing for differences in method of application, paint rheology and the application environment. New ships have several coats of paint, the number and composition of which do not vary greatly. The advent of Foul Release coatings, however, does not longer support his argument and in future a texture parameter will have to be included in the roughness characterisation if the added drag is to be predicted. This in turn requires the modification of the commercial version of the Hull Roughness Analyser. Roughness profiles are to be stored digitally. In order to calculate the spectral parameters and Fractal Dimension accurately by the acquisition of a sufficient number of data, a smaller sampling interval is also recommended. In order to validate any prediction method, the acquisition of full-scale data is ultimately

required and to the Authors knowledge this has not yet been done for hulls coated with Foul Release coatings. Foul Release surfaces, however, quickly acquire a slime film, which unlike other fouling organisms does not release when the vessel is underway. The added drag of a slime film compared to a newly applied coating is likely to be significant, but limited (i.e. restricted to a few percent) (Candries et al., 2002b). This research project is ongoing and water tunnel experiments are planned at the University of Newcastle-upon-Tyne to study the drag, boundary-layer and roughness characteristics of Foul Release surfaces which have been immersed in seawater for one year. References ITTC, 1990a, Report of the Resistance and Flow Committee, Proceedings of the 19th International Towing Tank Conference, Madrid, Spain, 16-22 September 1990, Vol. 1, pp. 62-64. ITTC, 1990b, Report of the Powering Performance Committee, Proceedings of the 19th International Towing Tank Conference, Madrid, Spain, 16-22 September 1990, Vol. 1, pp. 262-265. Townsin, R.L., 1990, The MC correlation allowance: the evidence re-examined and supplemented, Marine Roughness and Drag Workshop, RINA, London, UK, Paper 9. Townsin, R.L., and Dey, S.K., 1990, The correlation of roughness drag with surface characteristics, Marine Roughness and Drag Workshop, RINA, London, UK, Paper 8. Candries, M., 2001, Drag, boundary-layer and roughness characteristics of marine surfaces coated with antifoulings, PhD Thesis, Department of Marine Technology, University of Newcastle-upon-Tyne, UK.

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume III


Candries, M., Atlar, M., Guerrero, A., and Anderson, C.D., 2001, Lower frictional resistance characteristics of Foul Release systems, Proceedings of the Eight International Symposium on the Practical Design of Ships and Other Floating Structures (PRADS 2001), Elsevier, Amsterdam, The Netherlands, Vol. 1, pp. 517-523. Candries, M., Atlar, M., Mesbahl, E., and Pazouki, K., 2002a, The measurement of drag characteristics of Tin-free, SelfPolishing Co-polymers and Foul(ing) Release coatings using a rotor apparatus, Proceedings of the 11th International Congress on Marine Corrosion and Fouling, San Diego, CA, USA, 21-26 July 2002. Grigson, C.W.B., 1982, The drag coefficients of a range of ship surfaces, 11. Trans. RINA, Vol. 124, pp. 183-198. Townsin, R.L., Medhurst, J.S., Hamlin, N.A., and Sedat, B.S., 1984, Progress in calculating the resistance of ships with homogeneous or distributed roughness, N.E.C.I.E.S. Centenary Conference on Marine Propulsion, Paper 6. Candries, M., Atlar, M., and Anderson, C.D., 2002b, Estimating the impact of newgeneration antifoulings on ship performance: the presence of slime, Submitted to the Journal of Marine Engineering and Technology. I.4. Discussion on the Report of the 23rd ITTC Propulsion Committee: Comments on Powering Performance Prediction By: Michael Schmiechen, Germany The PC Report deals with the well known scale effects in model screw propeller performance essentially without drawing consequences. The usual way out is to perform

open water tests, even with wake adapted propellers, at sufficiently high Reynolds numbers. But in model propulsion tests the propellers are usually run at much lower Reynolds numbers, though in the behind condition. And the powering performance analysis is based on these two sets of incoherent data! This is the simple reason that since many years I am promoting the powering performance analysis solely based on propulsion tests, on model and full scale propeller performance in the behind conditions. Accordingly, in order to perform the powering performance analysis in the traditional conceptual framework, axioms, constitutive equations, have to be introduced for wake and thrust deduction, permitting the robust determination of their values. As it has been mentioned in the Session on Speed and Powering Trials this goal has finally been reached in the rational evaluation of quasisteady tests, model and full scale. In this connection I would further like to point out, that four of the traditional methods for evaluating traditional trials discussed by the SC on Speed and Powering Trials are based on model propeller open water tests. In view of the scale effects in propeller performance and wake, this is neither desirable nor necessary, as it has been shown. I wonder what the opinion of the Propulsion Committee is on these matters. I.5. Discussion on the Report of the 23rd ITTC Propulsion Committee: Use of New Antifouling Coatings By: Mehmet Atlar and Maxim Candries, University of Newcastle-upon-Tyne, United Kingdom Due to environmental side effect related with TBT, the I.M.O. has decided in October 2001 to prohibit the application of TBT-SPCs from 2003 and hence completely phase out


The Propulsion Committee

23rd International Towing Tank Conference

their use by 2008. One of the alternatives, which is biocide free, is Fouling Release coating system: These coatings are silicon elastomers, which have entirely different surface characteristics, providing low surface energy, so that firm attachment of fouling organisms is avoided and the release of the fouling organisms occurs at sufficiently high service speeds. Because these low energy surfaces also provide reduced skin friction drag characteristics, they have been recently applied on the propeller of large ocean-going tankers and container ships. There are reports on notable fuel saving after the applications claimed by respectable shipping companies and leading paint companies. Recently completed PhD research at Newcastle University involved an experimental investigation on the drag, boundary layer and roughness characteristic of Foul(ing) Release surface. This investigations confirmed the dray reduction characteristics of this paint (Candries, 2001; Candries et al., 2001; Candries et al., 2002). As continuation of this project a new set of experimental research is underway to investigate the performance of these paints on propellers involving full-scale trials (Atlar et al., 2002). One of the inherent difficulties encountered with these paint systems is the analysis of their different texture characteristics. Whereas the TBT or TBT free SPC (Self Polishing Copolymers) displays a typical closed texture and hence reasonable correlations with roughness height, the Fouling Release system exhibits a wavy open texture and does not correlate with a single roughness height requiring further parameters representing its texture (Candries & Atlar, 2002). Investigations are underway at Newcastle University to establish such parameters and the surface characteristics of propellers coated with fouling release to drag coefficient of the propeller. This will have impact on the performance

characteristics of the propeller through their analysis. The main objective of this contribution is to report on newly emerging means to improve the propeller performance as well as to protect them against fouling. It will be also useful the Propulsion Committee to think about how these surfaces could be modelled and their effects be taken into account for performance analysis. References Candries, M., 2001, Drag, boundary-layer and roughness characteristics of marine surfaces coated with antifoulings, PhD Thesis, Department of Marine Technology, University of Newcastle-upon-Tyne, UK. Candries, M., Atlar, M., Guerrero, A., and Anderson, C.D., 2001, Lower frictional resistance characteristics of Foul Release systems, Proceedings of the Eight International Symposium on the Practical Design of Ships and Other Floating Structures (PRADS 2001), Elsevier, Amsterdam, The Netherlands, Vol. 1, pp. 517-523. Candries, M., Atlar, M., Mesbahl, E., and Pazouki, K., 2002, The measurement of drag characteristics of Tin-free, SelfPolishing Co-polymers and Foul(ing) Release coatings using a rotor apparatus, Proceedings of the 11th International Congress on Marine Corrosion and Fouling, San Diego, CA, USA, 21-26 July 2002. Atlar, M., Glover, E.J., Candries, M., Mutton, R.J., and Anderson, C.D., 2002, The effect of a fouling release coating on propeller performance, 2nd Conference on Marine Science and Technology on Environmental Sustainability (ENSUS 2002), University of Newcastle-upon-Tyne, UK.

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume III


Candries, M., and Atlar, M., 2002, Reconsideration of the correlation of roughness and drag characteristics of surface coated with anti-fouling, 23rd ITTC, Discussion to Resistance Committee, Venice, Italy. I.6. Discussion on the Report of the 23rd ITTC Propulsion Committee: Some comments on the Chapters Form Factor Prediction from The Gothenburg 2000 Workshop and Development of the Formulation of the Flat Friction Line By: I.A. Chicherin and A.V. Pustoshny, KSRI, St. Petersburg, Russia Based on RANS-code calculations carried out in different laboratories worldwide, some comments have arisen after the analyses of the results of the form-factor for the tanker KVLCC2. In the Table presented by the Propulsion Committee, form-factor was determined based on the ITTC-57 flat plate friction line. Table I.6.1 below makes a more comprehensive analysis of RANS-code applicability to full-scale resistance prediction. To determine the form-factor presented in Table I.6.1, the Grigson line was applied, a recommendation from the Committee for further investigation was presented. The form-factor determined was based on results of RANS code calculation of viscous resistance (columns 12 and 13). One can find that the calculated hull form-factors for the model and full scale conditions are practically the same (ignoring the values themselves as the gap between components of viscous resistance from different calculations were quite big). For all options of flat plate friction line the gap between model and full scale form factors are rather big. These results are not accidental; they may be confirmed by the results of Ishikawa

(1996) for the tanker "Ryuko-maru as well as experience of RANScode calculations in KSRI. Table I.6.2 contains the same analysis for two hulls Bulker and Cargo; flow separation for both hulls in model scale was found. The Bulker impact of separation on resistance was small (residual resistance coefficient CR=0.610-3 at Fr=0.1), the Cargo impact of separation on resistance was rather high (residual resistance coefficient -3 CR=1.410 at Fr=0.1). As it is clear from Table I.6.2, for the Bulker, the tendency was the same as in calculations for KVLCC2. For hull option Cargo, due to the fact that because of separation the auto-model correlation is not quite correct, form-factor for model and full scale are different. The conclusion is that auto-model correlation of form-factor takes place if form-factor is determined based on calculated viscous resistance of the hull. It is possible to scale viscous resistance by form-factor; but simultaneously, the results demonstrated that flat plate friction lines at least are not universal. If we take this conclusion, the question arises of how everyone manages to predict, so accurately, full-scale propulsion by the ITTC78 procedure. The answer may be found in Sames (1996) calculations performed for Hamburg Test Case (HTC) at full scale Re=1.2109. These results coincided with those of scaling by the ITTC-78 when this procedure was performed to full scale, i.e. when correlation factor CA=0.410-3 was implemented. This means that implementation of a new friction line should be done with great caution because such implementation will require changes in the system of correlation factors. Returning to RANS codes, it should be taken into account that RANS codes at the moment are not so accurate for precise computational prediction of form-factor. On the other hand, based on RANS-code calculations, it is possible to develop the procedures for viscous resistance scaling without determining the


The Propulsion Committee

23rd International Towing Tank Conference

form-factor, because form-factor determination is always a procedure with high elements of art and judgement. If we use the form-factor formula CV,S = (1+k)CF0,S for initial determination of form-factor, we can obtain:

CV ,S = CV ,m

CF0 ,S CF0 ,m

As disadvantages we see the complication in unification of procedures from all societies of researchers and the need for development of a new system of correlation factors. It is possible to note here that the ITTC-78 is applied in different centers with some variations. The implementation of a new flat plate friction line will require a new system of correlation factors. In conclusion, the consideration of the implementation of a new flat plate friction line (i.e. Grigsons frictional formula) is a rather late problem especially taking into consideration that 5-10 years is required for implementation of a new friction line and all hydrodynamic centers will have their own preference. Commercial RANS-code use and accuracy of calculation undoubtedly will increase. References Ishikawa, S., 1996, Study on Scale Effect on Viscous Flow around Hull and on Propulsive Performance of a Ship by Using CFD, Transaction of the West-Japan Society of Naval Architects, No. 91, March 1996, pp. 1-14. Sames, P., 1996, Resistance and Wake Prediction by Computing Turbulent Ship Flows, Ship Technology Research, Vol. 43, No. 3, August 1996, pp. 124-136.

In the last formula it is possible to determine the proportional coefficient by the relation of calculated values of viscous resistance of the hull for model and full scale conditions instead of the relation of viscous resistance of a flat plate for model and full scale Re. Therefore, the formula for scaling of viscous resistance will be

CV ,S = CV ,m

calc CV ,S calc CV ,m

In Table I.6.1 the proportional coefficient is presented in column 14. The average value of this coefficient is equal to 0.463, standard deviation 0.033, coefficient of variation 7%; therefore, the scatter is not so high from one calculation to another. The main advantages of such procedures of viscous resistance scaling are exclusive of form-factor determined by experiment and exclude limitations from the character of the hull flow.

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume III


Table I.6.1


Model scale
CTm103 CFm103

Full scale
CTs103 CFs103

k(model) k(ship)

CF0 Grigson
k(model) k(ship)

CF Calculated
k(model) k(ship) CTs/CTm


8 0.273

10 0.341


12 0.276




4.392 3.441

Horus-easm 4.460 3.630 2.110 1.750 0.293 0.502 0.361 0.416 0.229 0.206 0.473 Horus-rij-w 4.230 3.300 1.820 1.470 0.226 0.295 0.291 0.221 0.282 0.238 0.430 Horus-sst

4.700 3.920 2.230 1.920 0.362 0.587 0.435 0.497 0.199 0.161 0.474 4.059 3.357 1.766 1.461 0.177 0.257 0.239 0.185 0.209 0.209 0.435

5 FLUENT Fluent 6

MARINParnassos 4.323 3.870 1.850 1.685 0.253 0.317 0.320 0.242 0.117 0.098 0.428 IST Neptune Surf CFX


4.090 3.320 4.210 3.370

0.186 0.220

0.248 0.285

0.232 0.249

4.329 3.397 1.934 1.551 0.255 0.377 0.321 0.298 0.274 0.247 0.447 4.340 2.910 2.236 1.940 0.258 0.591 0.325 0.501 0.491 0.153 0.515 4.660 3.590 4.320 3.560 Grigson CF0(model)=3.27610e-3 CF0(ship)=1.49010e-3 k = CT/CF0(Grigson) 1 0.351 0.252 0.422 0.319 0.298 0.213 3.886 3.361 1.944 1.578 0.126 0.384 0.186 0.305 0.156 0.232 0.500


ITTC-57 CF0(model) = 3.4510-3 CF0(ship) = 1.40510-3 k = CT/CF0(ITTC) 1 Table I.6.2

Model scale
CTm103 CFm103

Calculated CF(model) = column(5) CF(ship) = column(7) k = CT/CF(Calculated) 1

Full scale
CTs103 CFs103

k(model) CTm103

CF0 Grigson
CFm103 CTs103

CF Calculated
CFs103 k(model)

1 Bulker Cargo

2 3.81 4.29

3 2.92 2.97

4 2.17 2.58

5 1.65 1.55

6 0.224 0.334

7 0.302 0.596

8 0.259 0.384

9 0.238 0.515

10 0.305 0.444

11 0.315 0.665


The Propulsion Committee

23rd International Towing Tank Conference

II. COMMITTEE REPLIES II.1. Reply of the 23rd ITTC Propulsion Committee to A.F. Molland The Propulsion Committee thanks Dr. Molland for his questions about the form factors of high speed vessels. We fully agree that formally a value of 1+k different from 1 would be required to preserve consistency within the extrapolation method. However, the progressively increasing uncertainty of the experimental determination of the form factor with increasing transom area and the observation that reasonable predictions can be made either assuming 1+k = 1, or using a statistical method for 1+k, has led to the acceptance of alternative procedures for high speed ships in which other adaptations are being made as a varied wetted surface area. The committee recognises the inconsistency. Nevertheless, the committee is not in a position to promote form factor techniques which, though physically more correct, will not find much enthusiasm due to a lack of practical applicability to this class of ships. This expectation is based on the persistent reluctance observed in several institutes to use form factors, even for slow and medium speed merchant ships. II.2. Reply of the 23rd ITTC Propulsion Committee to Shaoxin Wang The Propulsion Committee thanks Mr. Shaoxin Wang for his contribution. The original wake scaling formula in the ITTC 1978 method was not really intended, nor suited to accurately predict the full scale effective wake fraction of twin screw ships and of very slender single screw ships as well. The new proposed formula could be a good way out within the scope of the ITTC 1978 procedure. However, the problem is then shifted to the selection of the parameter q. In q the cumulative

effects of the rudders behind the propellers, the effect of the pre-rotation, the sense of the propeller rotation and changed local flow velocities must be present. The committee invites Mr. Shaoxin Wang to provide guidance how q should be chosen in dependence of the factors mentioned. II.3. Reply of the 23rd ITTC Propulsion Committee to M. Candries and M. Atlar The Propulsion Committee wishes to thank Dr. Candries for a further detailed discussion on foul release coatings as they compare to modern anti-fouling coatings. As proposed at the discussions end, in-service performance degradation over time will be of primary concern. The occurrence of slime coating, etc, requires careful at-sea surface simulation, which will be difficult to maintain in the laboratory environment. Coating degradation well beyond one year immersion would also be of interest, along with the quantification of surface degradation with mechanical cleaning. The committee hopes that this area of research can continue and model scale to full scale correlation data for these new coatings can be accumulated. This would hopefully lead to improved prediction of in-service powering estimates along with the improved performance these coating provide. II.4. Reply of the 23rd ITTC Propulsion Committee to M. Schmiechen The committee thanks Dr. Schmiechen for stimulating the on-going discussion to rationalise the model tests, the analysis of the results of the experiments and the analysis of and interpretation of full scale data. The committee agrees that in several procedures there are some inconsistances, such as open water tests on wake adapted propellers, ap-

23rd International Towing Tank Conference

Proceedings of the 23rd ITTC Volume III


plying data at different Reynolds numbers and other comparisons of incompatible data. Nevertheless, the tests as carried out and analysed are to provide ship builders, owners, designers and propeller manufacturers with data within a familiar framework. Systematic deviations in scientific consistency of the procedures are accepted. Wake fractions, thrust figures, power and rotation rate predictions gain in reliability if they are determined within the usual framework. The committee fears that Dr. Schmiechens long-term struggle to convince the ITTC community that a totally new approach would be superior will continue as too drastic changes are proposed at one time. The new definitions of look-alike propulsion parameters, even using the same terminology, will not easily become accepted if there is the slightest doubt of incompatibility with the traditionally used data. The committee suggests that the modernisation in this area of the profession should be carried out step by step, first showing that within the traditional framework significant gains in understanding and clarity can be attained incrementally, rather than by rebuilding the whole classical structure of experimental propulsion research. II.5. Reply of the 23rd ITTC Propulsion Committee to M. Atlar and M. Candries The propulsion committee wishes to thank Prof. Atlar on an interesting discussion on the use of fouling release coating systems. Clearly, these systems offer potential improvements in drag reduction, and a departure

from the performance of traditional coating systems. This new coating system will offer a challenge on interpretation of model to full scale correlation, and possibly reconsideration of required powering allowances. We hope that in the future, Prof. Atlar and his colleagues can provide the research necessary to help address these issues. II.6. Reply of the 23rd ITTC Propulsion Committee to L.A. Chicherin and A.V. Pustoshny The Propulsion comittee wishes to thank Dr.s Chicherin and Pustoshny for an illuminating discussion into the future direction of powering prediction using CFD. We agree with the authors comments concerning form factor and ability to compute k factor. From that standpoint, application of CFD needs significant work. The authors observed a reasonable consistency in the ratio of CT computed at model and full scale Re. This result shows great promise. Ultimately, this is the quantity of interest for the prediction of full scale performance from model test measurements. The authors should be applauded for their insight into the future use of CFD. Perhaps we are closer than we think in incorporating a viscous, computational based procedure to replace Froudes original hypothesis. To get to a universial use of CFD in this regard will require continued development in the verification and validation of computational methods, and finally a code certification process that is agreed upon by the ITTC community.