Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
24th Symposium on Naval Hydrodynamics Fukuoka, JAPAN, 8-13 July 2002 Propeller Inflow at Full Scale During a Manoeuvre G.Kuiperi, M.Grimm2, B. McNeice2, D.Noble3, M.Krikke4 (~ Marlin, The Netheriands, 2Royal Australian Navy, 3DRDC Atiantic Canada, 4Royal Netheriands Navy) ABSTRACT The propeller inflow of a patrol boat of the Australian Navy was measured at full scale using a newly developed LDV technique. The velocity distribution in front of the propeller was measured, both on a straight course and in a turn. Model tests that were carried out in a towing tank with the model at a drift angle were shown to adequately simulate the most important effects of the wake distribution. The model wake data were used to complete the full scale data and an analysis was made of the power absorption of the propellers in relation to the inflow velocities. It was found that the transverse velocities in the propeller plane were almost fully responsible for changes in power absorption in a turn. The mean axial velocity was about equal for the inner and the outer propeller in a turn. Unexpectedly the transverse velocities induced by a turn were small in the upper half of the propeller plane. The effects of the inflow measurements on the cavitation inception speed have also been estimated analytically. INTRODUCTION Propeller design has traditionally been based on the propeller inflow on a straight course under ideal conditions. In operational conditions effects of waves, ship motions and manoeuvres can however be important. This is especially so for the cavitation inception speed of a naval vessel because of the increase of radiated noise when cavitation starts. Optimization of the cavitation inception speed on a straight course may lead to a propeller which has a high inception speed, but which is very sensitive to small disturbances in the propeller inflow. In practice the inception speed will be much lower than in ideal conditions on a straight course. The effects on the propeller inflow of waves, ship motions and manoeuvres are still not predictable. The decrease of the cavitation inception speed of Navy ships in a seaway is often obvious. This is of course partly due to increases in the propeller loading caused by added resistance in waves and partly due to wave velocities and the resulting ship motions. Observations of cavitation in heavy seas also indicate that the effect of the rudder in coursekeeping can be important. In some cases the autopilot was switched off in heavy seas and the fluctuations in cavitation decreased significantly. The effect of the rudder can also be experienced when monitoring vibrations due to cavitation in a seaway. The vibrations come and go, apparently without any relation with the ship motions. The rudder actions, required for coursekeeping, are then the cause of the variations in vibrations. In the present project the focus was primarily on the effects of manoeuvres on the inflow of the propeller, although some attention was also paid to the ship motions. Since cavitation inception is of primary importance for Navy ships, the focus was also on frigate type, twin screw configurations. FULL SCALE MEASUREMENTS Since very little is known about the effects of manoeuvres on propeller inflow and cavitation, this study can be characterized as exploratory. Investigations performed at model scale would have to be extrapolated to full scale. However, the differences between model and full scale wake cannot yet be fully described analytically, particularly when subject to ship motions in a seaway or while manocuvring. This means that, to arrive at conclusions for full scale from model tests, empirical corrections would have to be applied. Consequently it was decided to focus on full scale measurements and cavitation observations. Model tests were carried out
after the full scale tests in a preliminary effort to simulate the full scale phenomena at model scale. During the period 1991-1996 the Canadian Navy, the Royal Australian Navy, the Dutch Navy and Marin jointly collaborated in conducting full scale sea trials to investigate the effects of operational conditions on propeller cavitation inception. These trials included wake field measurements and cavitation observations on the Canadian Forces Auxiliary Vessel CFA V QUEST and on the Australian patrol boat HMAS TOVVNSVILLE. The wake measurements on CFA V QUEST did not cover the whole propeller disk reliably, but the measurements on HMAS TOVVNSVILLE produced a complete dataset over a significant portion of the propeller disk. Therefore, in this paper the results of the measurements on HMAS TOVVNSVILLE obtained in July 1993 are reported and analyzed. These trials consisted of Laser Doppler Velocimetry (LDV) measurements of the inflow into one main propeller for both straight line running and moderate turns to both port and starboard. The LDV measurements were complemented by cavitation observations and video recording as well as shaft speed and torque measurements.The intention was for these trials to be undertaken in calm water such that only the effects of steady manoeuvring were observed. The triple screw patrol boat of the Royal Australian Navy on which measurements were performed is shown in Figure 1. Figure 1: The Australian patrol boat HMAS TOWNSVILLE The main particulars of the ship and outer propellers are given in Table 1. Two outboard fixed pitch propellers are the main propulsion while the smaller auxiliary, controllable pitch propeller at the centerline is only intended for slow speed operations and as such is feathered when the main propellers are in operation. The propeller arrangement is shown in Figures 2 and 3. The outer propellers have their shafts supported by A-brackets and turn outward over the top. Figure 2 also shows the relative locations of the LDV head, its measuring plane and video camera used on the sea trial. Waterline length Beam Draft Displacement Propeller diameter I Propeller pitch at 0.7R Number of blades Expanded Blade Area Ratio 38.5 m 6.54 m 2.145 m 25 1.3 tons 1.275 m 1.168m 3 1.221 Table 1: Main particulars of IMPS TOWNSVILLE. ll ~~ 1< 1 ~~ AUX~ ~WE}[rSP4f:E _~ ~ - WE Figure 2: Position of LDV head, measuring plane and video camera housing relative to A-bracket struts and main propeller Figure 3: Shaft arrangement showing both main propellers and the center propeller in the feathering position. 2
A skeg at the center line of the boat extends aft to within 20% of the waterline length forward of the transom, which is slightly forward of the position where the propeller shaft enters the hull. The centerline auxiliary propeller was in the feathered position during all measurements. The drift angles involved in the present investigations were limited to approximately 5 degrees, in order to minimize the effect of the presence of the center propeller on the propeller inflow of either main propeller. MEASURING EQUIPMENT A special LDV head was developed for this project to be able to measure flow velocities up to a distance of about 4 meters from the head. A cross section of the head is given in Figure 4. After splitting the beam both beams are led along the side of the head using fixed mirrors. The last mirror is controllable, so that the angle of the beam can be adjusted. This makes it possible to locate the measuring volume from between about 20cm from the head out to 4 meters. Proper crossing of both beams is of course extremely important. Following the movement of air bubbles or impurities in the water through the control volume, the scattered light is transmitted through the window and focussed by two mirrors, a large one and a small one. It is led into a glass here in the center of the large mirror for further processing. The proper focussing of the light on the glass fibre is important and requires a high accuracy of the mirrors. The signal from the glass fibre is transformed into an electrical signal and processed by Dantec Burst Spectrum Analyzers (BSA). ~ WA - t j .. , , ~1~1l~ _ \ ~ / _ I Hi \ 1 / I CO~Iq=LA~,t PI - *$ Figure 4: LDV head For use in sea trials the head of the LDV system was reduced to a tube of 45 cm in length and 25 cm in diameter. This made it possible to mount the head above a window in the hull. The space between the head and the window was filled with distilled water to avoid breaking the path of the laser beams. As indicated in Figure 5 the maximum angle between the perpendicular to the window and the axis of the head was 30 degrees. This allowed the major portion of the inflow ahead of the propeller disk to be measured, as illustrated in Figure 7. Both the head of the LDV system and its laser unit were located in the steering gear compartment of the patrol boat. The 5 Watt laser, required to obtain a proper signal up to 4 meters outboard of the hull, was large (length about 1.2 meters), heavy and required a large external water cooling unit, which was placed on the main deck. This made it impossible to connect the laser directly to the head. The laser was therefore installed separately at a distance of about one meter from the window with the head. Mirrors were used to bring the laser beam into the head. In this set-up it was crucial that the laser beam entered the head along the centerline. This was done using a proper set-up of the mirrors. However, vibrations of the ship made the beam unsteady relative to the head. This was overcome by actively controlling the mirrors on the head in such a way that relative motions between the laser and the head were compensated. The active control was done by reading a part of the laser beam through the mirror on a Charge Coupled Device (CCD). The active mirror was then adjusted for deviations of the beam from the required position. It has to be mentioned that present day lasers with equal power are small enough to be coupled directly to the laser head. ,~4 / I _OPTK.:~" I ~ lit ~ salts ~ 1 1 ' 1 1~ ifs 1 Figure 5: Position of optical head above hull window 3
Due to the acute angle of the intersecting pair of laser beams, the measuring volume at a distance of 3.5 meters outboard of the hull was approximately 10 cm long. Close to the hull this length reduced to less than 1 cm. The width of the measuring volume was always less than 5 mm. Processing software analysed a burst from particles or bubbles passing the measuring volume and checked whether the shape of the burst was acceptable. Only then was the burst stored. The ratio between the accepted and the total burst rate is called the acceptance ratio. In ideal conditions the acceptance ratio can be as high as 70 percent. During the sea trial the acceptance ratio varied from a few percent up to 25 percent, depending on the distance to the measuring volume. In a turn the acceptance ratio was much lower than on a straight course, indicating that increased turbulence in the flow can complicate the LDV measurements. Observations of cavitation on the port propeller were made using the Osprey video camera illustrated in Figure 6. A clear hemispherical dome at one end of the camera contained the adjustable optics. The camera dome was fitted inside a larger acrylic dome that extended a short distance outside the hull. For the sea trials on HMAS TOWNSVILLE, the camera was located between the A-bracket arms just forward of the LDV window. The camera was remotely controlled from inside the ship to provide significant pan, tilt and rotation of the view during observations. To freeze the images of the individual blades in successive revolutions, strobe light units synchronized to the propeller rpm were placed on adjacent windows. FW4CIE ~ CAMERA rem _ Be_ Figure 6: Video Camera Installation The pulsed signals generated from a digital magnetic sensor as it detected the teeth on a wheel around the propeller shaft were processed through a specially designed viewing controller to allow time- lapsed observations of the blades at fixed angular positions (for further details see Kennedy et al. 1989~. Additionally the propeller torque and rpm and the rudder angle were measured. LDV MEASUREMENTS AT FULL SCALE Measurements of the propeller inflow were carried out on the port propeller. The measurement grid is given in Figure 7. The measuring plane was upstream of the propeller, as shown in Figure 2. The zero degree radial is the vertical through the propeller shaft centerline. The radials at which full scale measurements were carried out are from -30 (inboard) to +30 (outboard) degrees, with steps of 5 degrees. Because of the shaft bracket bossing the radials from -5 to +5 degrees could only be measured above the shaft. The distance of the measuring points, measured from the center of the optical head, was from 0.25 m. to 1.75 m. in steps of 0.25 m. The center ofthe optical head was 18.75 cm from the hull, so the inner points of the grid were just outside the hull. The measuring plane was upstream of the propeller, as given in Figure 2. Figure 7: The measuring grid. The open locations were not measured at full scale, the black locations were not measured at model scale . The LDV head at full scale allows measurements in one direction only. So only the velocity components in the plane perpendicular to the laser beams could be measured. The direction perpendicular to the measuring plane will be called 4
the axial direction. It coincides approximately with the axial direction of the ship. The direction along the laser beam radials in Fig. 7 will be called the "vertical" direction. The velocities in this direction could not be measured at full scale. The direction along the arcs in Fig. 7 will be called the "horizontal" component. The vertical and horizontal components of the transverse wake field can then be translated into the common tangential and radial velocity components in the cylindrical coordinate system of the propeller. Since the horizontal velocity components could be negative, a Bragg cell was initially used in the LDV head. The Bragg cell gives a phase shift to the laser beam, causing the interference grid in the measuring volume to move at a certain speed. However, the Bragg cell caused a decrease in beam intensity and was therefore omitted. To determine the direction of the "horizontal" velocities in the measuring plane, i.e. whether flowing inboard or outboard, two measurements were taken at every measuring point in directions 30 degrees on each side of the axial direction. The horizontal and axial velocity components were derived from these two components. The vertical component could not be measured with this system. With the exception of measurement points far from the hull each LDV measurement at full scale had a maximum duration of one minute. The measurement was completed in less time if 3000 bursts were recorded. The number of bursts during one measurement varied significantly. The maximum burst rate was found close to the hull. An example is shown in Figure 8a. There was considerable scatter in the velocities due to turbulence in the ship's boundary layer. This scatter decreased considerably at greater distances from the hull, as shown in Figure 8b. At the maximum distance of 1.75 meters the maximum duration of the measurements was therefore set at 3 minutes instead of 1 minute for all other locations. Even then only a limited number of bursts (from 2 to 10) could be measured. This was apparently due to a lack of seeding, although occasionally large increases in the burst rate was measured, probably caused by patches of bubble clouds passing through the measurement volume. No efforts were made to apply artificial seeding. l . :: ~ 0 ~ ~ I' ~ I'.$ velocity mlsec a L count ... .. ;2- , . ,! ... , ,, Celtic ~~ . . '.' b Figure 8. Recording of axial velocities at 25 cm distance (close to the hull, Figure a) and at 125 cm distance from the hull (Figure b) on a straight course TEST CONDITIONS AT FULL SCALE Extreme clarity of the water is required for sufficient backscatter signal strength at full scale. LDV measurements were carried out in the waters near Cairns (Australia) during winter, where the water conditions are very good. The measurements were done inside the Great Barrier Reefs. Moderate winds of 15 knots with waves approximately one meter in height were present during the measurements on a straight course and at the end of the measurements in a turn to port. The weather conditions were better during the turns to starboard and during the cavitation observations. There was considerable current in this location. To investigate the effects of a turn, the propeller inflow velocities were measured in the axial direction on a straight course and in two directions (resulting in the axial and horizontal components of the inflow) in a turn. The turn was set at a fixed rudder angle of 8 degrees with both propellers nominally at 520 rpm. The position of the ship was measured using a GPS reading at every 45 degrees of heading. The result for a turn to starboard is given in Figure 9. An average drift can be obtained from the shift of locations at the same heading. By canceling out the average drift the resulting path becomes as shown in Figure 10. From Figure 10 an average turning diameter was determined as well as an average drift angle. The diameter was found from the distance between opposite locations on the path and this varied with the heading. The average turning diameter in Figure 10 is 785 meters. The drift angle was found from the deviation of the heading (0, 45, ...270 and 305 degrees) and from the tangent to the turning circle. 5
This tangent is perpendicular to the lines connecting opposite points on the path in Figure 10. In", 8;X ~ 9= ~ m~m ~ "W' Figure 9: Path of the ship with rudder 8 degrees to starboard. ~ _ I ~ ~ i_ -" - t a'/ I I I -3~ ~ 1~ 1~ ~~ 1~ 1~ 1~ 1~ t~ 320D ~ - i +2 / E _ At= Figure 10: Path corrected for drift with rudder 8 degrees to starboard. The drift angles varied between 1 and 5 degrees. The path of the ship was slightly elliptical instead of circular, due to the wind. The wind also caused the variation in drift angle over the turn, that increased with the wind strength. It turned out that the turning diameter was difficult to reproduce, even at nominally the same rudder angle. Therefore the measurement of the turning circle had to be repeated every time the rudder was reset. The time required to complete a full circle was also measured. In combination with the turning radius this gave the average ship speed in a turn. The speed in a starboard turn was thus estimated at 12.4 knots. On a straight course, also with a shaft speed of 520 rpm, the speed was 13 knots. The LDV measurements took several days of running in circles. During the measurements it was occasionally necessary to interrupt the turns and continue measurements at another location. The turning diameters and ship speeds varied during the duration of the measurements, partly due to the lack of repeatability of the rudder angle, partly due to changes in weather conditions, especially the wind. During the last part of the measurements in a turn to port the wind had increased considerably and it was necessary to increase the rudder angle to 13 degrees in order to turn the ship properly. The turning diameter varied between 670 and 915 meters, the estimated ship speed during the turns varied between 12.4 and 12.9 knots and the drift angles varied between -1 and +5 degrees. These variations in conditions are incorporated in the measured wake field, as the time to measure the full wake field was many hours and the trial spanned over several days There is no reliable method to correct each inflow reading for the effects of wind and wave induced ship motions. THE FULL SCALE WAKE FIELD On a straight course only the axial wake component was measured. The results are shown in Figure 11. The interpolation between the measured points is taken linearly, to avoid artificial peaks and valleys in the velocity contours. It should be mentioned that in this and the following wake distributions the propeller disk was projected on the wake field along the propeller shaft. This resulted in the propeller disk being slightly further upwards in the wake field than when it was projected along the baseline, as in Figure 7. The effect of the propeller position in the wake field is considered small. In Figure 11 the non-dimensional velocity (l-w) of the outer flow is close to unity, indicating that the ship speed of 13 knots, which was used to derive the non-dimensional velocities, was correct. Since the velocities are those of the total wake, including the induced velocities of the propeller, the occurrence of a velocity of 1.08 in the center of the propeller is perfectly plausible. In the top position two wake peaks showing a minimum velocity (1-w) of 0.7 are present. The ridge in between these peaks (1-w=0.94) is supported only by one point. Although the scatter at this point was considerable (it is the point given in Figure 8a) it seems reliable. 6
indicating that in this turn the reference velocity was higher than 12.4 knots, as it was in the turn to port. .. Figure 11: Non-dimensional axial velocities on a straight course. In a turn two velocity components were measured, which were decomposed into the axial and the horizontal velocity components. The axial velocity field in a turn to port is given in Figure 12. Since the measurements were taken on the port propeller, this is the inner propeller in a turn. The velocity at the outer measurements arc is around 1.09, indicating that the estimated ship speed of 12.4 knots, based on the turning diameter and time to turn) was underestimated. This is supported by the axial wake of the propeller in a turn to starboard. Taking this into account, the wake velocity in the top portion of the disk again has two minima, providing wake peaks of 0.92 (1.01 from Figure 12 minus 0.09) and 0.94 (1.03 minus 0.09~. While the axial wake peak of the inner propeller in the top position has widened and has become less deep due to the turn it has remained in the same general location. Low velocities of 0.94 (1.03 minus 0.09) are also seen at two locations along the left (outboard) side of the propeller disk in Figure 12. The horizontal velocity component in a turn to port is given in Figure 13. The most significant aspect of this result is that this velocity component is much smaller in the upper half of the propeller disk than in the lower half. The consequences of this feature will be analyzed below. The non-dimensional axial velocity distribution for the total flow into the outer propeller in a turn to starboard is given in Figure 14. Again the outer flow has a non-dimensional velocity of 1.08, 7 i ..0. / '~ i/> l. ~- .1 lit me. ! - .' / 'i ~ ,'L/ `,T,~0 Figure 12: Non-dimensional axial velocities in a turn to port (inner propeller) EB to CL Figure 13: Horizontal velocity component in a turn to port (inner propeller, ship speed 6.39 m/see) Taking this into account the two axial wake peaks of the outer propeller are 0.89 (0.97 from Figure 14 minus 0.08) and 0.77 (0.85 minus 0.08). The minimum velocity in the upper part of the outer propeller is also increased (from 0.7 on a straight course), but not as much as in the wake peak of the
inner propeller. Figure 14 also shows some widening of the wake peak relative to the straight course. Figure 14: Non-dimensional axial velocity distribution in a turn to starboard (outer propeller) The horizontal velocity component in a turn to starboard, so for the outer propeller, is given in Figure 15. Again there is a stronger horizontal velocity component in the lower half of the propeller disk, as was the case for the inner propeller, but now especially in the outer sector. To analyze these measured propeller inflow data further the powering data of the propellers at full scale were measured. In combination with model test results, as will be given below, these powering data can be related to the propeller inflow. POWER ABSORPTION AT FULL SCALE Extensive torque/rpm measurements were done both on a straight course and in a turn. On a straight course the power absorption at 520 rpm was 400 kW per shaft and the difference between the shafts was negligible. In a turn the shaft torque and rpm measurements were taken with the rudder at 8 and 12 degrees. The results are given in Table 2. Table 2 shows that the inner propeller was overloaded while in a turn. The variation of the overload reflects the lack of repeatability of the turning radius, which was not measured during these tests. Port S3 ~ 340 Figure 15: Horizontal velocity component in a turn to starboard (outer propeller,ship speed 6.39 m/sec). Trial condition straight 8 deg. SB rudder 1 2 deg. SB rudder 8 deg. Port rudder 12 deg. Port rudder Outer propeller _ shaft _ = Pow Port . SB SB pm 520 518 521 520 519 Inner propeller KW 400 400 404 400 400 shaft . SB SB Port Port rpm 520 519 520 521 522 Table 2: Shaft power measurements in a turn. kW 400 422 460 453 481 For subsequent model tests with 5 degrees of drift angle the average conditions from table 2 for port and starboard turns at 8 degrees rudder angle were taken: 520 rpm and an average power absorption of 400 kW of the outer propeller and 436 kW of the inner propeller. It may be noted that the rpm of both propellers in a turn remained very similar to the straight run values. This is considered to be related to the engine governing system on HM245 TOVVNSVILLE. LDV MEASUREMENTS AT MODEL SCALE For an analysis of the full scale data a comparison with model test data is necessary. These model test data will also be used to estimate the vertical velocity component at full scale. This component is required to be able to do propeller calculations for the inflow at full scale. These propeller calculations will be used 8
to relate the measured powering characteristics with the propeller inflow and to reach conclusions. Proper simulation of turns at model scale would best be accomplished using a rotating arm facility, but no such facility was available. If it can be assumed that the major effect of a turn is due to the drift angle, model tests performed in a regular towing tank can be used. This was done for a model of HM24S TOWNSVILLE with scale ratio 8. As mentioned above the drift angle was taken as 5 degrees, a representative drift angle for the condition of 8 degrees rudder. The model speed without drift was scaled from 13 knots to 2.36 m/sec. The propeller rotation rate of the model was chosen such that the torque coefficient KQ both at model and full scale was 0.028. This ignores the viscous correction on torque and thrust due to the difference in Reynolds number between model and full scale, but this effect is small. The resulting model rpm was 1368 rpm. With S degrees drift angle the ship speed of approximately 12.4 knots was scaled to a model speed of 2.24 m/sec. In this condition the rpm of the model was proportionally reduced from the zero drift case to 1300 rpm. This means that the advance ratio of the propeller was kept the same with and without drift angle to simulate the same rpm in a turn and on a straight course at full scale At model scale 3-D LDV measurements were made at the same grid locations as were used at full scale. In the model tests the LDV head was on one side of the model and the shadow of the shaft bracket bossing now blocked measurements in the inboard part of the grid. The density of the measurements was less than at full scale. The measured locations are indicated in Figure 17. The axial velocity distribution of the model with no drift and with the propeller working is given in Figure 16. The transverse velocity distribution is given in Figure 17. (For the model scale data the propeller disk was projected on the measuring plane along the baseline, resulting is a slightly lower position of the indicated disk than at full scale. The measured points are at the same locations, however). The torque coefficient of both propellers was 0.028, the same as at full scale. Comparison of the axial wake peak with the one at full scale (Figure 11) shows that the double wake peak is also present at model scale, but the peaks are less pronounced, and wider apart. This is connected to the hull boundary layer that at full scale is much closer to the hull. ~ ~ "J _ - ~/ ,2;~ f~,'./,Jn. ~ __ ~ ~ ~ - ~ 3 \ ~ /1 ~- \ I. ~ __ 3 f7 / 1 ,~' `~', f Figure 16: Non-dimensional axial velocity distribution on the model without drift. 1 imps] / 1 ~ 4~' ~ _ ~ _ / rat ~ 7` ~ ~ 1., \\ / / /! Figure 17: Transverse velocity distribution on the model without drift (model scale velocities, model speed 2.36 m/sec). The velocities at model scale are more uniform, especially above the propeller shaft. Propeller calculations will be used later to assess the effect of these differences on the inception speed. The transverse velocities at model scale cannot be compared to full scale because the latter were not measured. Figure 17 indicates that the 9
"horizontal" velocity component is small over the whole disk. The circulation in the inflow is very small, as can be expected for this type of hullform. The axial and transverse velocities with a drift of 5 degrees are given in Figures 18 and 19 for the inner propeller and in Figures 20 and 21 for the outer propeller. The torque coefficient of the inner propeller was 0.031 and that of the outer propeller was 0.026. The drift angle therefore caused a reduction of the loading of the outer propeller and an increase of the loading of the inner propeller. In a turn at full scale the propeller loading of the inner propeller was increased to 0.031, but the loading of the outer propeller was unaffected. The relation between the wake distributions and the propeller performance will be further analyzed below using lifting surface calculations. The axial velocity distribution in front of the inner propeller with the model at 5 degrees drift is shown in Figure 18. The velocities still have two minima, as for the model at zero drift. The wake peaks are shifted slightly inboard instead of outboard and the inboard peak was not fully measured. The depth of the wake peak was not much different from zero drift. ~ , Jr) . , J I ~ f Aim; a' -' . ,f ~ W~ . ~ Figure 18: Non dimensional axial velocity distribution of the inner propeller behind the model with 5 degrees drift angle (model speed2.24 m/sec). ~ ail \\ \ _~ _ am_ - , ` - ~ \ \\ , . 1/ t~ ~~ ~ ~ ~ / F\ ~ $W Fs ~ ~ W f ~~W Figure 19: Transverse velocity distribution of the inner propeller behind the model with 5 degrees drift angle (model speed 2.24 m/sec). The transverse velocity of the inner propeller at 5 degrees drift is given in Figure 19. Similar to the observations in a turn at full scale (Figure 13) the transverse velocities in the upper part of the propeller disk are small. There is a clockwise swirl in the transverse velocity. The magnitude of the transverse velocity will be examined later. , ~ ~ 1 1 ,tJ ~ it, ~ .~.- ~ _ ~.~1 Figure 20: Non dimensional axial velocity distribution of the outer propeller behind the model with 5 degrees drift angle (model speed 2.24 m/sec). 10
between the horizontal velocities at model and full scale are given in Figure 22. The grid was interpolated to equal grid points both for the model and the ship. ~ _ - ) ) fly v~ l ~ \ t 2 ~ t f ~ _~. / ~ ~ ~ ,- , ~ j ., W~ ',' ~~; ~ ~ ~ // Figure 21: Transverse velocity distribution of the outer propeller behind the model with 5 degrees drift angle (model speed 2.24 m/see) The velocity fields of the outer propeller with the model at 5 degrees drift are shown in Figures 20 and 21. A comparison with the model results without drift (Figures 16 and 17) shows that the wake peak in the upper part of the propeller shifted further inwards, which corresponds with higher horizontal velocity components just above the shaft (Figure 21~. The horizontal velocities closer to the hull become small again. COMPLETION OF THE FULL SCALE VELOCITY DISTRIBUTION. For an analysis of the full scale wake data it is necessary to estimate the vertical velocity component. For this the model data were used. It is possible to derive the vertical velocity component from the two other measured components using the continuity equation in combination with the assumption of zero axial gradient of the velocity. However, this attempt failed due to the limited accuracy of the measured data. A simpler approach was therefore used. Assuming that the measured wake distributions of the model at 5 degrees drift angle are representative of the wake in a turn at full scale, the vertical velocity component at model scale can be used to complete the full scale data. The validity of this assumption can be assessed by comparing the horizontal velocity components measured on the model and the ship. The outer horizontal components of the inner propeller at model scale (Figure 19), non-dimensionalized by the model speed, are typically 0. 15. The difference Figure 22: Difference in non-dimensional horizontal velocity component of the inner propeller between model tests at 5 degrees drift and the ship in a turn. = Y.l al ./ 1 I' 43.H Figure 23: Completed non-dimensional transverse velocity field of the inner propeller of the ship in a turn. 11
In the region where the data are measured both at model and full scale the maximum differences are approximately 0.1, but in the lower half of the propeller disk these differences are generally smaller. This illustrates the feasibility of using the model values of the vertical velocity component to complement the full scale wake. The inaccuracy of the vertical velocity component is the more acceptable here since this velocity component generally plays only a minor role in regions where cavitation inception is expected to occur (the top part of the disk and, in a turn, the bottom part of the disk). - . //! . : - 63.75 O. 1 Figure 24: Completed non-dimensional transverse velocity field of the outer propeller of the ship in a turn. Adding in the vertical component from the model wake, the completed transverse velocity fields at full scale are given in Figures 23 and 24. These velocity fields, in combination with the measured axial velocity distributions, can now be used for calculations of cavitation inception on the propeller and for relating changes in the wake field to the power data. ESTIMATION OF THE TRANSVERSE VELOCITIES An important result from both the model and full scale measurements was that the transverse velocities above the propeller shaft were very small, despite the fact that the skeg ended at 20% of the ship's length upstream of the rudder. The third propeller, especially in feathered position, may have been a factor here. The small transverse velocities in the upper half of the propeller were confirmed at model scale, where the center propeller was also present. This showed that the drift angle was a major factor for the propeller inflow in a turn and not the rate of rotation. The undisturbed horizontal velocity component Vh due to a drift angle at ship speed Vs is Vh/Vs =sin 6, where ~ is the drift angle. With a drift angle of 5 degrees this component is VhtVs =0.087. For a ship in a turn the rotation of the ship causes an additional horizontal velocity component in the propeller disk. Estimating the distance to the ship's center of rotation from the propeller plane to be 3/4L, where L is the ship length, and the time to complete a full circle as 7 minutes, the horizontal velocity component due to the turn is Vh/Vs =0.068. The undisturbed horizontal velocities due to a turn and due to a drift angle are therefore of the same order of magnitude. This was not always true for the measurements made at model scale and full scale. Figures 25 and 26 show the maximum transverse velocities divided by the speed of the model or ship as a function of the non- dimensional radius in the propeller disk. 0.3 no VtransNs AL,_,: .. i_ . ~ ~ 1 l ~ I , ,' I ~ . ' 1 ~ 0.20 0.50 0.80 1.0 RlRmax '. ~ 1, Figure 25: Maximum transverse velocities on the inner propeller. VtranstVs 00~7¢ `` ~ ~ 0.10 ~ ~ ~ 408 -- t I -- ~ 0.20 0.50 RlRmax 0.80 1.0 Figure 26: Maximum transverse velocities on the outer propeller. 12
At inner radii there is a discrepancy, but at the outer radii the transverse velocities of the ship in a turn and the model at a drift angle are almost equal. For both the inner and outer propellers on the ship the maximum transverse velocity occurs in the lower half of the propeller disk. Although the undisturbed horizontal velocities are almost equally caused by drift and rotation, Figures 25 and 26 show that the transverse velocity in the propeller disk is dominated by the drift angle. Model tests with a drift angle can therefore simulate a ship in a turn. On the other hand the drift angle during a turn is a major factor in changing the propeller inflow. The maximum transverse velocity was between 1.5 and 2 times the horizontal velocity induced by the drift angle. This is much higher than would be found from potential flow calculations around the hull. The viscous region above the propeller increases the transverse velocities in the lower half of the propeller disk and strongly decreases the transverse velocities in the upper half of the propeller disk. This has consequences for the propeller loading, which will be analyzed next. TOTAL WAKE DISTRIBUTION AND PROPELLER LOADING The measured model and full scale wake distributions can be used to calculate the performance of the propeller in terms of propulsive performance (thrust and torque) and minimum pressure. The latter is of course important for cavitation inception. The important role of the transverse velocity field made it impossible to analyze the propeller performance in terms of Taylor wake fractions, using the open water characteristics of the propeller. Instead lifting surface calculations were carried out with the Marin program ANPRO. The velocity distribution of the measured total wake, made non- dimensional with the ship or model speed, was used as input for the wake distribution. The mean axial velocity was adjusted to obtain the measured torque coefficient. The result was a pressure distribution on the propeller (including the minimum pressure) and an effective mean axial velocity Ve. This effective axial velocity, derived from the required torque coefficient, can be compared with the LDV measurements. Since the ship or model speed was known, this mean axial velocity was written as a wake fraction 1-w~VJVs. This is a calculated wake fraction which includes the effects of the transverse velocity. For the transverse wake distribution of the ship, the model data used for the straight course and for only the vertical component in turns. The result of these calculations is summarized in Table 3. It is noteworthy that the presently calculated wake fractions are equal at full scale both on a straight course and in a turn and, for the model, with and without drift. This leads to the conclusion that the axial inflow velocity of the propeller is not changed by the turn or drift angle. The difference in power absorption between port and starboard propellers is therefore fully due to the transverse velocity field. Condition KQ 1-w 1-w calculated measured LDV Ship straight 0.028 1.03 1.03 Ship in turn 0.031 1.07~1.01) 1.11~1.05) to port Ship in turn 0.028 1.05 (0.99) 1.11 (1.05) to starboard Model 0.028 0.96 1.02 straight Model drift 0.031 0.94 1.04 to port Model drift 0.026 0.96 1.05 to starboard Table 3: Calculated wake fractions using lifting surface calculations (the data in brackets are for the ship speed in a turn of 13 knots instead of 12.4 knots) In Table 3 the calculated wake fractions of the ship are greater than one, which is not plausible. As indicated by the velocities in the outer flow, obtained from the LDV measurements, the average ship speed in a turn was probably higher than was calculated from the transit time of the turn. Setting the ship speed in a turn equal to the ship speed on a straight course (13 knots) lowers the calculated wake fractions of the ship in a turn to values close to one and the measured wake fractions of the LDV data in Table 3 to around 1.05. An estimate of the induced velocities at the distance of the measuring plane upstream of the propeller indicates that the propeller induced velocities there are between 5 and 10 percent. Therefore, except for the ship wake on a straight course, the results in Table 6 are consistent. The calculated wake fraction of 1.03 for the ship on a straight course seems approximately 5 percent too high. This might be caused by the fact that the transverse velocities of the model were used for this calculation. 13
CAVITATION OBSERVATIONS The location of the camera was between the shaft brackets, as shown in Figure 2.0nly the upper half of the propeller blades could be observed from that position, so the observations were limited to the top half of the propeller disk. From the calculations it was found that the cavitation extent did not change very much from 20 degrees before until 20 degrees beyond the top position of the blades. This was also the range where the largest effect on cavitation was expected. The cavitation pattern on a straight course with 520 rpm is shown in Figure 27. The front blade is near the top position. A small amount of sheet cavitation is visible. A cavitating tip vortex is still visible at about 120 degrees on the blade just ahead. In a turn to starboard the outer propeller is observed and the cavitation pattern does not change significantly (Figure 28~. In a turn to port (Figure 29) the propeller is overloaded, but this leads to only to a slight increase of the cavitation, particularly further inward from the tip. The overloading is mainly in the lower half of the propeller disk and thus does not affect the cavitation pattern in the top position of the blades. These observations were confirmed by calculations of the cavity extent at the full scale torque coefficient. On a straight course the calculated cavitation was incipient near 0.95R, but hardly visible. In a turn, both to port and to starboard, a slight amount of sheet cavitation was calculated (Figure 30~. So the effect of a turn on the cavity extent was small. 14 Figure 27: Cavitation on the ship's port propeller at 520 rpm on a straight course Figure 28: Cavitation on the ship's port propeller at 520 rpm in a turn to starboard (outer propeller) Figure 29: Cavitation on the ship's port propeller at 520 rpm in a turn to port (inner propeller) s.T \ ~ / , i. NS 1 ,/ ~ ~ . '' _ _ I I I I I I I I ~ I r ~ I . -. -~.28 B.~8 8.08 0.49 Figure 30: Calculated cavitation on the propeller in the ship's wake in a turn to port..
The model tests showed hardly any cavitation near the propeller tip, indicating that part of the full scale cavitation was tip vortex cavitation, which is delayed at model scale. In order to observe the differences between the cavitation patterns in a drift to port and a drift to starboard, the propeller was overloaded by increasing the rpm by 10%. (Figures Stand 32~. These predictions showed that the main difference in the cavitation pattern occurred in the downgoing blade (9Odegrees and further), where the transverse flow began to influence the blade loading. The differences in the top position of the blade were small. Figure 31: Observed cavitation on the model with 5 degrees drift to port (Vm=2.234 m/sec., nm-1426 rpm). Figure 32: Observed cavitation on the model with 5 degrees drift to starboard (Vm-2.234 m/sec., nm-1426 rpm). Therefore, contrary to what was expected, cavitation in the upper half of the propeller disk was rather insensitive to manoeuvres. This can be explained by the small effect of manoeuvres on the propeller inflow in the top position. The main effect of a manouevre, at least in this case, is in the lower half of the propeller disk. This insight can be used to assess the effects of higher rudder angles on cavitation inception. EFFECTS OF A TURN ON THE INCEPTION SPEED Assuming a linear relation between the axial inflow in a turn and on a straight course, the inflow distribution at various turning angles can be found from a linear interpolation or extrapolation of the measured wakefields. A similar linear relationship is assumed between the torque coefficient on a straight course and in a turn. With these assumptions the wake fields have been calculated from a straight course up to 1.5 times the turning rate used on HMAS TOWNSVILLE in steps of 0.25. The turning rate is thus expressed as a fraction of the full scale experiment. These wakefields have been used as input for lifting surface calculations to find the minimum pressure on the blades over one revolution. When the minimum pressure is below the vapor pressure, cavitation is expected. Assuming that the propeller advance ratio, and thus the pressure distribution, does not change when the rpm is reduced, we can find the reduction of the rpm required to make the minimum pressure equal to the vapor pressure. The advance ratio then gives the inception speed at the reduced rpm. In Figure 33 the inception speed of both the inner and outer propeller is given as a function of the turning rate. Initially, at lower rates of turn, the inception speed increases. This is due to the reduction of the wake peak behind the shaft and brackets in a turn. However, as the turning rate increases the transverse velocity becomes increasingly important, and above 0.5 times the turning rate of the full scale test, the inception speed continuously decreases. The minimum pressure under these conditions does not occur when the blade is in the top position, but when it is in the lower half of the propeller disk, where the larger transverse velocities have increased the blade loading and decreased the pressure. At 1.5 times the 15
turning rate of the full scale experiment, the inception speed has dropped more than 1.2 mYsec on the inner propeller. Inception Speed versus turn strength 5.0 1 1 4.0 ~1 Mater ropeler . . ~ inrurprup 1 1 ! 000 02S 050 075 IM. Or Figure 33: Inception speed of the inner and outer propeller as a function of the turning rate. (A turning rate of 1 is the turning rate of the full scale trial. The vertical grid distance is 0.2 m/sec). Figure 33 shows a rather complicated behaviour for the changes to the inception speed. Initially the outer propeller is the critical propeller. With increasing rate of turn, however, the inner propeller takes over. These predictions were not verified by inception tests at full scale. CONCLUSIONS Contrary to what is generally expected the full scale experiment on HMAS TOVVNSVILLE shows that the main effect of a turn on the propeller inflow is in the transverse velocities! The axial inflow of both propellers in a turn remained the same. The differences in propeller loading could be explained fully by the transverse velocities. This explains why on frigates with inward turning propellers the inner propeller is unloaded in a turn. According to the full scale measurements, the transverse velocities in the upper half of the propeller disk were small. The presence of the center propeller may have blocked the transverse flow. This blocking was not unfavourable, because in a turn the wake peak due to the shaft and the struts was less concentrated and thus less deep. The combination of a small transverse velocity and a wake peak reduction caused the inception speed to increase initially at small rates of turn. For fuller ships the transverse velocity in the upper part of the propeller disk may not be negligible. The velocity will then be inwards, towards the ships' centerline. For cavitation inception it is then favourable to have the propellers turning inward over the top, and use the unloading effect of the transverse velocities to offset the loading effect of the axial wake peak. It is remarkable that the properties of the inflow in a turn could be simulated with a model at a drift angle. The results show that the drift angle, and not the rate of rotation in a turn, is important for the crossflow in the propeller plane. This is important also when the ship is on a straight course in seaway, where the mean rate of turn is zero, but where significant drift angles can occur due to coursekeeping by the rudder. The effect of rudder actions on the inception speed requires further investigation. The crossflow in the top position in a turn was also investigated on CFAV QUEST. This is a different type of ship, with bossings instead of brackets. Here the effects of a turn on the inception speed were stronger. The data are, however, less complete than on HM245 TOVVNSVILLE. The collaboration between the original participating organizations in this project is continuing. The conduct of full-scale trials on CFAV QUEST, HMAS TOWNSVILLE and the Canadian frigate HMCS NIPIGON and associated model tests have been completed. The focus of the project has now turned to the development of analytical tools and methodologies for the prediction of inflow and propeller cavitation performance. In parallel, the feasibility of implementing alternative operating strategies for a naval ship to increase its cavitation inception speed is to be examined as an analytical exercise. Such measures include for example an adjustment of propeller pitch setting to suit the prevailing operating conditions in the case of frigates equipped with controllable pitch propellers. Design guidance for consideration of propeller inflow conditions under realistic operational conditions are also expected to flow from the experience gained from this project. REFERENCES Kennedy, J.L., Sponagle, N.C., Wheaton, D.W., MacDonald, P. and Creaser, R.W., "Video Systems for Propeller Viewing Trials", Proceedings of the 22n~ American Towing Tank Conference St. John's , , Newfoundland, 1989, pp. 197-202 16
