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.
24eh Symposium on Naval Hydrodynamics Fukuoka, JAPAN, ~13 July 2002 Microbubbles: Drag Reduction Mechanism and Applicability to Ships Yoshiaki Kodama~, Akira Kakugawa~, Takahito Takahashi~, Shigeki Na- gaya~, K~7~,v~, C,,olv~m~ CR97Q Committee of the. <;hinh,~ildin~ Re- . . . .. I ~~ ~ ~~ berm = ~~~ search Association of Japan. (iNational Maritime Research Institute, Japan) Abstract Measurements on the skin friction reduction effect by microbubbles were reviewed. The data by several investigators were re-plotted using air layer thickness ta and compared among them, showing the differ- ences in air injection methods such as porous plates and array-of-holes plates, test section dimensions and the degree of wall effect. The mechanism for the skin friction reduction was briefly discussed, including the density effect, bubble size effect, and the reduction of the Reynolds stress. Some factors related to the ap- plication of microbubbles to full-scale ships were dis- cussed, and it was shown that net power saving is ~t- tainable using existing techniques. Finally, a full-scale microbubble experiment using a 116m-long ship was shown, in which 3% drag reduction was obtained, re- sulting in 2% net power saving. l.Introduction Microbubbles, i.e. small bubbles injected into the turbulent boundary layer on a solid wall, reduce skin friction significantly. Studies on microbubbles started by a pioneering work by McCormick and Bhat- tacharyya (1973~. The significant skin friction reduc- tion effect of microbubbles makes them look attractive for the military application, where techniques to realize significant skin friction reduction are regarded as in- dispensable for very high-speed ships or torpedoes. Therefore, in the age of the Cold War, their studies were mainly carried out in the United States and the former Soviet Union. In the 1970s, microbubble studies were carried out intensively in the former Soviet Union, e.g. Bogdevich (1977~. In the 1980s, the intensive study was carried out this time in the United States, headed by a group it the Pennsylvania State University. In the 16th Symposium on Naval Hydrodynamics Merkle et al. (1987) gave interesting observations on the rela- tion between the bubble distribution in the boundary layer and the skin friction reduction effect, and Merkle and Deutch (1990) reviewed the history of microbubble studies and described in detail the research activities at Penn. State. In the l990s, the center of the study moved to Ja- pan, aiming at their application to commercial ships. Ships that carry heavy loads such as crude oil, ore and grain, play an important role in the worldwide trans- portation. Their characteristics are that they are very large and that they move very slowly. Among the two major drag components of such ships, the free-surface wave component, being proportional to the square of the ship speed, is very small due to their low speed. Therefore the skin frictional drag component occupies approximately 80% of the total drag, resulting in its reduction being very important. Figure 2 shows an image of the application of microbubbles to a tanker ship. The University of Tokyo started the microbub- ble study in Japan. They carried out both experimen- tal and theoretical studies, jointly in some part with IHI (Kato et al. 1994, Guin et al. 1996, Kato et al. 1998, Yoshida et al. 1998, Watanabe et al. 19981. In the Na- tional Maritime Research Institute, Japan (NMRI), formerly the Ship Research Institute, the microbubble studies were carried out, first on the basic characteris- tics (Takahashi 1997, Kodama 2000), and then on the scale effect using a SOm-long flat plate ship (Takahashi 20011. Computational studies were also carried out (Kawamura 2001~. The purpose of the present paper is to discuss m- crobubbles for the drag reduction mechanism and the applicability to full-scale ships. In the 2nd chapter the skin friction mechanism is discussed. In the 3rd chapter the applicability of mic~bubbles to ships is discussed. In the fourth chapter a brief outline of the - ~ I I 1111, 1 1, 1. ~11! ill 1 _ 1 _ ~ _ _~! Figure 1 An image of microbubbles applied to a tanker ship
2 full-scale microbubble experiment carried out last year in Japan is described. 2. Skin friction reduction by microbubbles In this chapter we discuss how much microbubbles reduce skin friction in detail, mainly by viewing the authors experiment using a small circulating water tun- nel (Kodama 20001. As shown in Fig.l, the skin friction reduction ~- fect of microbubbles increases as the amount of i~- jected air volume ncreases, and reaches 80%, but it should be noted that the measurement was made at only 50mm to 65mm downstream of the point of bubble injection. As shown in Fig.l, in the practical applica- tion of microbubbles it is natural to assume that bubbles are injected at a small portion of the hull surface, to minimize installation cost and fouling possibility, and move along the surface thus covering a wide area effi- ciently. Therefore, for practical purposes, it is impor- tant not only to find out how much the bubbles reduce skin friction but also to find out how long the skin fric- tion reduction effect persists in the downstream direc- tion from the point of injection. 2.1 Experiment at The Pennsylvania State Univ. In the beginning a standard microbubble experi- mental data is shown. Madavan et al. (1984) carried out microbubble experiments using a water tunnel with a test section of BC x HC x LC = 114mm X508mmX760mm size ("c" for "channel"~. They used the top or bottom wall for air injection and skin friction measurement. Fig.2 shows the device in case the bottom wall was used for the experiment. Air was injected through a sintered metal porous plate (PP here- after) of Ba X Ha =102mm X 178mm size ("a" for "air") and 5 Am nominal pore diameter. The skin now CON nova UPPER Was Em=_ \TUNNEL L - IS, ~~ - .~ .~ · . ~rr~i It Il ~ ~ 1 . by Sll~Ea ~~1 ~" ~ POROUS PLAN ~ "~t ~ - 41N GAUGE - ~ ON LEG "S. IN ~" HE "91U tWA - -FILLY) Figure 2 The layout for gas injection and skin friction measurement (plate-on-bottom) (Madavan et al. 1984) friction was measured using a drag balance of B X L=102mm X 254mm size located immediately downstream of the air injection plate, insulting that the center of the drag balance is 266mm downstream of the center of the air injection plate. Figure 3 shows the measured skin friction reduc- tion effect in case the air injection and skin friction measurement were carried out at the top wall (plate-on-top). The horizontal axis shows the air n- jection rate Qa non-dimensionalized by the injection area Sa and the free-stream speed U,~, and the ver- tical axis shows the ratio of the measured skin friction in the bubble condition to that in the non-bubble condi- tion. The skin friction reduction data at various flow speed collapses nearly to a single line, and increases as the air injection rate increases, reaching 80% at the maximum. Since Merkle and Deutsch 1990 experimentally showed that the length Pa of the injection plate has little influence on the skin friction reduction, the use of the injection area Sa to nondimensionalize Qa has little physical meaning,. Instead, they suggested to use Cv, the ratio of volumetric flo w rates CV -Q QQ (1) a w 0.4 0.2 ~ It,& ~ 4. 2m/s · 9. 3m/s 0 12. 4m/s 0 17.4m/s o . . . . . . . 1 0 0.02 0.04 0.06 0.08 QISU sir ~ Figure 3 Skin friction reduction by microbubbles (Madavan et al. 1984) 2
3 where QW is the volumetric water flow rate in the boundary layer. But we have shown (see Fig.25) that the boundary layer thickness has little significance, and therefore, following Fukuda et al. 1999, prefer the use of the dimensional air flow thickness ta pa_ Qa (mm) (2) In Fig.3 the corresponding ta horizontal axis is also shown. 2.2 Kawakita's experiment in a cavitation tunnel (l)Array-of-holes plate (AMP) Although a porous plate is the most popular means for air injection, it has some problems. One is that the actual size of the holes is unknown except for the pore size shown in a catalog. The other is the non-uniformity of generated bubbles. The bubbles are observed to be generated not uniformly over the surface but at limited number of points distributed non-uniformly. The third problem is that the pressure loss across the plate is significant, i.e. 0.2 to 0.5 kgf/cm2 depending on the injection rate. The fourth is possible fouling at sea. The former two problems cause the repeatability of experiments questionable, and the latter two make the practical use of a porous plate difficult. Instead of a porous plate, Kodama et al. 2000 manufactured a plate with holes of lmm diameter with 3mm spanwise Smm streamwise pitches, and named it an array-of-holes plate (AHP hereafter). Kawakita and Takano 2000 carried out mic~bubble experiments in cavitation tunnel using an array-of-holes plate of the same type. Their experiments will now be bdescribed in detail. (2)Test facility Fig. 4 shows the layout of the test section. The dimensions of the test section are BC x HC = 500mm X500mm. On the upper wall they put a flat plate of 300mm width, 2150mm length upstream of the injec- tion plate for growing boundary layer and 1 500m length downstream for the measurement. The distance between the upper wall and the bottom side, i.e. the measurement side, of the flat plate was 180mm, result- ing that the distance between the flat plate and the lower wall was 320mm, which can be regarded as large enough to make the wall effect in bubble conditions negligible. The boundary layer thickness in the meas- urement locations were 25 to 40mm, according to the LDV measurement. Air was injected through an array-of-holes plate of Ba x L`a =237mmX 20mm. Skin friction sensors (for detail, see 2.3) were placed at xa =0.Sm, 1.0m, and 1.5m from the center of the injection plate, along the line 75mm from the center line, in order to avoid the wake of the propeller dynamorreter. ~: = .~ ~si ~ ~ ~ ~_e ~ -k ~ ~ Figure 4 Layout of test apparatus of Kawakita's periment in their cavitation tunnel (Kawakita and Takano 2000) (3) Cf / Cfo of AHP They plotted the measured Cf /Cfo as a function of CX, the average void ratio in the boundary layer _ _ Qa Q +Q a w (3) where the boundary layer thickness was estimated us- ing an empirical formula. CX is the same as Cv shown in eq.~31. OC can be converted to ta . Using eqs. (2) and (3), and their definition of the boundary layer thickness, it results ta = 0.122xRex-l/7 <4 ~ _ 1r' ~ v~ where x=2.15m. The results are shown in Fig.5. In all the results, the skin friction reduction is almost lin- ear with ta except for the small plateau region at small ta values, and it seems the plateau region be- 3
4 comes larger at higher speeds. At U=Sm/s, there is no clear difference among the data at three xa locations, but at U=7m/s and 1 Om/s the reduction becomes smaller at larger xa locations. In contrast to the independence of the skin friction reduction effect on flow speed in the experiments by Merkle and Deutsch 1990, their results show strong dependence on the flow speed, with smaller reduction at higher speeds. 1.1 1 o.s S 0.8 can 0.7 0.6 0.5 1.1 8 ~ - OXa=O.Sm ~ Xa=1 . Om XXa=1. 5m .................. . O 0.5 1 ~ (m m ) (a) U=Sm/s 1.5 2 ~08 . REX cow . 0.7 - ° Xa=0 . 5m to Xa= 1 . Om 0.6 XXa=1. 5m 0.5 . . . . . . . . . . . . . . . 0 0.5 1 0.9 0.8 cam 0.7 0.6 (b) U=7m/s ~ , ~ - OXa=0.5m ~ Xa=1 . Om XXPI=l .5m 1 1.5 2 t a(m m ) 0 0.5 1 t_a (mm) (c) U=lOm/s 1.5 2 Figure S Skin friction reduction by microbubbles (Ka- wakita and Takano 2000) (4) Cf / Cfo of PP and AHP It is interesting to compare the skin friction reduc- tion effect by a porous plate (PP) and that by an ~- ray-of-holes plate (AHP). Fig.6 shows the compari- son of the reduction using PP at xa=0.27m and at U=4.2m/s and 9.3m/s (Madavan et al. 1984) and that using AHP at xa =O.Sm and at U=Sm/s and lOm/s (Kawakita and Takano 2000). In spite of slight dif- ference in xa and U. they agree with each other well, which shows that the skin friction reduction effect by the two plates are nearly the same, at least at a point immediately downstream of bubble injection. 0.8 0.6 cam 0.4 0.2 . . - 0& to AMadavan PP 4.2m/s Xa=0.27m O O Madavan PP 9. 3m/s Xa=0 . 27m AKawaki ta AHP 5m/s Xa=0 . 5m X Kook i to AllP 1 Om/.c x~=n .Sm 0 1 2 3 t_a (mm) 4 5 Figure 6 Comparison of skin friction reduction using PP (Madavan 1984) and AHP (Kawakita and Ta- kano 2000) 2.3 Experiment at NMRI using Small High-speed Water Tunnel (HSWT) (l )Test facility In order to measure the skin friction reduction effect and its persistence in the downstream direction, the authors built a tunnel for microbubble experiments as shown in Fig. 7. (Kodama 2000y, The tunnel was designed following the design by Guin et al. 1996. Bubbles are injected at the air injec- tion chamber in the test section and removed by buoy- ancy in the dump tank to allow continuous operation. The test section of Be x HC x LC = lOOmm X 15mm X 3000mm size has been designed to be nearly two-dimensional in the transverse plane in order to avoid the side wall effect and be consistent with other test facilities for basic viscous flow experiment, and to be long enough in the streamwise direction to measure 4
the persistence of the skin friction reduction effect. The maximum flow speed at the test section is 12m/sec The air injection chamber is located 1038mm down- stream of the test section inlet, to have fully developed flow at the point of injection. This location is called Position 1 ~ xa =Om), followed by Positions 2 ~ xa =O.Sm), Position 3 (xa =O.Sm), and Position 4 (xa=l.Sm), where various measurement was carried out. Due to the narrow test section and high speed, the pressure gradient in the streamwise direction is not small, e.g. 0.16 atm/m at U=7m/s. Therefore the local pressure at each measurement location was taken into account to determine Qa. Hereafter the flow speed U denotes the bulk or average flow speed in the test section. Flow direction Air injection chamber (Posinonl) . .'. . . '. %, P2 P3 PA ~=e~e~ it ~ 1 net son son son 46i,1 Wire meshes 3000mm F Test sector ( Is loo ) ~ ~ . - ~' n mm x mm Electro-magnetic DOw meter Pump Figure 7 Small High-speed Water Tunnel at NMRI for microbubble experiment (2)Air injection Air is injected either through a porous plate (PP) made of sintered bronze with nominal pore udius of 2 ,u m shown in Figs. 8 and 9, or through an ~- ray-of-holes plate (AMP) shown in Fig. 10. The di- ameter of the holes is lmm and they have 3mm span- wise pitch and 5mm streamwise pitch. Both PP and AHP have the injection area of Ba x La = 72mm X 72mm. Air Porous Pant| - ~1 ~ 1 1.1. 1.1 v Flow ~ ~ , _ - -l _~~G_ _ ~ __ __ __ · an_ ~ ~ ~~ ~_~ __ _~9 _ ~~ 1 _"~" 1 _-,,. 1 _ - ~ ~ ~ 1 _ _ 1 . . (a)Section view (b)Test section with air injection chamber Figure 8 Air injection chamber and bubble generation through a porous plate Figure 9 Bubble generation from the porous plate (U=7m/sec) (3)Skin friction sensors Skin friction was measured using commercial skin friction sensors (Sankei Engineering co., ltd.) shown in Fig. 11. The sensor is a force gauge type (Kato 1994) with a sensing disk of lOmm diameter and the mam- mum load of 2 grams weight or 250 Nlm2. The sensors were placed along the center line of the test section. Static and dynamic calibration of the skin friction sensors was made (Kodama et al. 2000~. Figure 10 Array-of-holes plate $~2mm - Ammo / Sensor floating disk \\\\\\\\\\\\\ O O a O O O Flow ~ Figure 11 Skin friction sensor (4) Cfo correction for wall effect When air is injected in the test section, the water flow rate is kept constant, so the total flow speed in- creases due to the volumetric increase of the flow, and Cfo, the skin friction in the non-bubble condition, should also increase and therefore needs correction. Thus Cfo (Qa), i.e. Cfo in the bubble condition 5
6 was corrected using the following formulae. 1 1 Cfo (Qa ) = Cfo (O) [U(O)] (s) 09 ~ 0.8 where ~ 0.7 7(f ) = 0.03325pV l /4f 7 / 4r-l l 4 `6' 0.6 as U(Qa) Total flow speed at Qa. ° r: Hydraulic radius. The eq.~6) corresponds to the empirical Blasius formula (Schlichting 1968~. The above two equations result in on Cfo (Qa ~ ~U(Qa ~ ]7/4 (7) 0 0.8 Cfo (0) U(O) (a . ~ - +~ ARC (p 04< . . . . .' . a 2 3 t_a (mm) (a)U=5m/s . ~ ~ 00~ . °~b~: ~ ~ x and further, using equip), 0.6 ~ O ,~ ~ 0.5 , , to (1 ~ )~7/4 `8) ° t_a (mm) (b)U=7m/s is) Cf /Cfo of PP The skin friction reduction was measured at three speeds of U=5, 7, and lOm/sec, at xa =O.Sm (P2), l.Om (P3), and l.5m (Pip. The results are shown in Fig. 12. The results by Madavan 1984 at xa =0.27m are also shown. The dashed line shows the inverse of eq.~8), i.e. the contribution of the o.s 0.8 0.7 Cfo correction. 0.6 The Cf / Cfo values at three streamwise locations 05 are approximately the same, except that the values at xa =O.Sm are smaller in some cases, supporting the general understanding that the reduction effect becomes smaller in downstream. It may also be stated that the reduction becomes smaller at higher speed. The pre- sent skin friction reduction values are apparently smaller than those by Madavan 1984, whose reason is not clear. The contribution of the C correction is fo comparable to the reduction effect, and in many cases dominant. It is clear that the correction method needs further validation. O Xa=0.5m (P2) ~ Xa=l.Om (P3) X Xa=1.5m (P4) CfO (0) /CfO (Qa) ~ Madavan 4. 2m/s + Madavan 9. 3m/s O Xa=0.5m (P2) ~ Xa=l.Om (P3) X Xa=1.5m (P4) CfO(0) /CfO(Qa) O Madavan 4 . 2m/s ~ Madavan 9 . 3m/s MIX X Oo ~ o , . . . . ~ . 0 1 O Xa=0.5m (P2) · Xa=l.Om (P3) X Xa=1.5m (P4) C fO(O)/CfO(Qa) Madavan 4 . 2m/s Madavan 9 . 3m/s 2 3 t_a (mm) (c)U=lOm/s Figure 12 Skin friction reduction by microbubbles (NMRI HSWT porous plate) <6y Cf / Cfo of AHP The skin friction reduction results similar to Fig.12 but with the array-of-holes plate AHP is shown in Fig. 13. The overall tendency is similar to that of PP., but there are subtle differences. At large ta' the reduc- tion effect of AHP tends to saturate. The data scatter of AHP looks smaller than that of PP. 6
1 r no 08 n.6 0.9 0.8 0.7 0.6 0.5 0.9 0.8 0.7 0.6 0.5 _~ 0 1 2 3 t_a (mm) (a)U=Sm/s E`," °?~ t I ; . . . . ~ . (b)U=7m/ s t_a (mm) 3 . 0 1 2 3 t_a (mm) (c)U=lOm/s ° Xa=0.5m (P2) ~ Xa=l.Om (P3) X Xa=1.5m (P4) CfO (0) /CfO (Qa) Madavan 4. 2m/s t Madavan 9. 3m/s O Xa=0.5m (P2) · Xa=l.Om (P3) X Xa=1 . 5m (P4) CfO (0) /CfO (Qa) ~ Madavan 4. 2m/s + Madavan 9. 3m/s O Xa=0.5m (P2) · Xa=l.Om (P3) X Xa=1.5m (P4) CfO (0) /CfO (Qa) ~ Madavan 4 . 2m/s t Madavan 9 . 3m/s Figure 13 Skin friction reduction by microbubbles (NMRI HSWT array-of-holes plate) (7)Comparison with Madavan's result (PP) The skin friction reduction data of PP by Madavan (Fig.3) and NMRI (Fig.12) are compared, as shown in Fig.14. It is clear, first of all, that Madavan Ejected air much more than NMRI. The skin friction reduc- tion of NMRI is consistently smaller than that of Madavan. Since there is not much data in Madavan's result at ta less than two, no further comment is pos- 7 sible. 1.2 0.8 0.6 Cal 0.4 0.2 o . ~,~ <>Madavan Xa=0 . 27m 4. 2m/s Madavan Xa=0 . 27m 9. 3m/s ON M R I HSWT Xa=O. 5m 5m/s O NMRI HSWTXa=0.5m lOm/s ~S6~ O O 0 00 0 2 4 6 8 10 12 14 t_a (mm) Figure 14 Comparison of PP skin friction reduction by Madavan and NMRI (HSWT) (8)Comparison with Kawakita's result (AMP) The skin friction reduction data of AHP by Ka- wakita (Fig.5) and NMRI (Fig.13) are compared, as shown in Fig.15. They used the same type of AHP and measured skin friction at the same downstream locations using the same kind of sensors. At small air injection rate, i.e. at ta less than 1, the two results seem to agree well, but beyond that ta value NMRI values tend to saturate in spite of significant contribu- tion of the Cfo correction, while Kawakita's values still go down. This may suggest that at small air n- jection rate, where the wall effect is not significant, the phenomenon in a channel of small height is the same as that in a channel of large height, but at large air injec- tion rate, where the wall effect is significant, the phe- nomenon in a small channel becomes different from that in a large channel and further skin friction reduc- tion is prevented due to some unknown mechanisms. no 0.8 0.7 0.6 7 1 1 4) At\ ~ ~ ~,~ ' :'K . . . 0.5 0 1 2 3 t_a (mm) (a)U=Sm/s O NMRI Xa=0.5m ~ NMRI Xa=l.Om X NMRI Xa=1.5m CfO (0) /CfO (Qa) 0 Kawaki ta Xa=O. 5m Q Kawaki ta Xa=1. em OK Kawaki ta Xa=1 . 5m
8 in o.9 0.8 0.7 0.6 0.5 o 1 2 3 t_a (mm) (b)U=7m/s O NMRI Xa=O. 5m ~ NMRI Xa=l. Om X NMRI Xa=1. 5m CfO (0) /CfO (Qa) O Kawakita Xa=O. 5m Kawakita Xa=1. On Kawakita Xa=1.5m Figure 15 Comparison of AHP skin friction reduction by Kawakita and NMRI (HSWT) (9)Summary To summarize the comparison of the skin fric- tion reduction results from three test facilities, we cur- rently think as follows. e a) It is safer to use a large channel for microbubble experiment. b) In case a small channel is used, air injection rate should be limited. c) The Cfo correction method of eq. (5) needs further study for validation. d) The size of the cross section of a test section, not only the height just discussed but also the width, may have significant influence on the microbubble mechanism. The side wall influ- ence has not been examined so far. The influence of the width of an injection plate compared to the width of the channel has not been examined so far either. There are two factors that may contribute to the influence. One is the diffusion of injected bubbles into the bubble-free region. The other is the interaction of bubbles thus diffused with the boundary layer on the side wall. This influence may become more significant as the location goes further downstream from the injection point. The use of ta to compare results from different test facilities simply neglects this possible influence. 2.4 Experiments using 20m and 40m Hat plate ships by Watanabe 1998 From the practical application point of view, scale effect is the most important factor in the skin friction reduction effect by microbubbles. More specifically, we want to know how long the reduction effect of n- jected bubbles persists in the downstream direction. However, most of existing microbubble experiments are related to circulating water tunnels whose test sec- tion length is of the order of a few meters at most. So we have to do something to bridge the large gap be- tween a few meters of basic experiments and a few hundred meters of full-scale ships. Watanabe et al. 1998 carried out a pioneering ex- periment on the scale effect of microbubbles by towing a flat plate ship with maximum length of 40m at the maximum speed of 7m/s, corresponding to a typical cruising speed of large tankers, in their towing tank. (l)Tested ship and devices The tested ship had a long parallel middle part of 60cm width, plus streamlined bow and stern parts. The length of the bow part was 2.0m. The draft was 0.05m. The bow injection plate was located 1.2m from the bow end. It was a porous plate with 15 ,um pore diameter, and of size Ba x La = 250mm X70mm. For skin friction measurement they used commercial skin friction sensors of Sankei Engineering co., ltd., the same type as those used by MHI and NMRI, but with a larger sensing disk of 25 mm diame- ter and the ma~mum load of 10 grams weight. (2) Cf /Cfo The streamwise distribution of the measured local skin friction reduction at the ship speed V=7m/s is shown in Fig. 16, where the horizontal axis shows xa, the distance from the bow injection plate where air was injected, and the injected air rate is expressed with ta defined in eq.~21. It is seen that the reduction effect persists almost all the way up to the downstream end. 1.2 1 o 0.8 0.6 (J 0.4 0.2 0 0 n 0 ~ Q ~ ~ X X X X ° t_a=1.0mm ~ t_a=1.9mm x t a=2.9mm 0 10 20 30 40 Xa (m) Figure 16 Streamwise distribution of skin friction re- 8
9 auction of 40m-long flat plate ship (Fig.6 of Watanabe et al. 1998~. The same data is shown in Fig. 17 with ta in the horizontal axis. It is seen that the reduction increases non-linearly with ta. They also measured the total drag and its reduction. oo.8 0.6 0.4 0.2 x t ~ · Xa=1.8m · Xa=5.8m X Xa=13.8m Xa=21.8m Xa=37.8m 0 1 t_a (mm) 2 3 Figure 17 Skin friction reduction of 40m-long flat plate ship (the same data as those in Fig.16) 2.5 Experiments using a 12m flat plate ship at NMRI (l)Tested ship and devices Watanabe's experiment was carried out in a towing tank whose length was approximately 250m, and therefore at the speed of 7m/s the time for the meas- urement was very short. Also one cannot be sure whether the limited width of 60cm had significant ~n- fluence on the measured results or not. Therefore we decided to do a similar experiment but with the ship width of lm, in our 400m-long towing tank, and, as a first step, we made a 12m-long flat plate ship. It is shown in Figs. 18 and 19. The tested ship had a long parallel middle part of 60cm width, plus streamlined bow and stern parts. The length of the bow part was 3.0m and that of the stern part was 2.0m. The plan form of the two parts are, respectively (Bow part) (Stern part) y = 14 (3- x)4 - ~ 2 (3 - x)3 + 0.5 (9) y=-1x2+1x (10) 8 2 The bow part has the point of inflection at x= 1. Om, and the both parts connect smoothly to the parallel part. The draft was 0.045m. The corners were rounded with a bilge circle of 20mm radius. Turbulence stimulators of 2mm height were placed with tom pitch in the girthwise direction, at O.5m from the bow end. The bow injection plate was located 3.0m from the bow end. It was a porous plate with 2 An pore diameter, the same type as that used in the experiment shown in Section 2.3, and of size Ba x La = 500mm X lOOmm. For skin friction measurement we used commercial skin friction sensors of Sankei Engineering co., ltd., with a sensing disk of 25mm diameter, the same kind as that used by Watanabe. Bow trim guide Stern trim guide For~auge ~ TDwing carriage ~ ~L , 47 , ~ ~ w~tcr p~une Porous plate for air injection Skin faction sensors _ \ ~ / \ Flow controller ~ / / \ 0~: ~ im \ 81 ~ ~,Et°' ~11 33 ~ ~0 _ 1 _ _- . '1 1 2.000 mm Figure 18 12m-long flat plate ship of NMRI (PP) (a) Perspective view (b)Air injection plate (PP) Figure 19 Photographs of 12m flat plate ship of NMRI (2)Correction of skin friction values measured with lOgf sensors After the 12m-long ship experiment, we carried out the 50m-long ship experiment using the same skin friction sensors. In that case we repeated tests by ex- changing the sensors in order to find out data scatter due to different sensors. We found out that the meas- ured values by lOgf sensors were significantly higher than those by 2gf sensors, and that the latter were in good agreement with Schoenherr values. We sus- pected that the over-prediction by the lOgf sensors was caused by insufficient rigidity, and obtained correction 9
10 curves by fitting quadratic polynomials which go to unity at zero shear stress, to the data in bubble and non-bubble conditions, as shown in Fig.20, in which measurement locations are shown with the combination of the sensors used ("S" denotes lOgf full-scale sensors and "s" denotes lgf full-scale sensors). All the data shown in this section have thus been corrected. 1.2 It should be noted that the over-prediction of this kind causes over-prediction of skin friction reduction. So we asked the authors of Watanabe et al. 1988, who used the same kind of sensors of lOgf full scale, if they had similar experience, but the answer was that their measured shear stress was in good agreement with the Schoenherr value. The reason for the discrepancy is still unknown. 1.2 cc in O .v o 0.8 0.6 0.4 o 0.2 1.2 o 1.0 ~¢ 0.8 4~ _' 0.6 in, 0.4 a hi, 0.2 0.0 _ 0,o _ - - ~~o~ 0.0 0.2 0.4 0.6 t w o f lOgf (unit: gf/cm^2) (a) V=5m/s , W~ ~i,~ ~~ 0.0 0.S 1.0 1.5 t w 0 f lOgf (unit: gf/~2) (b)V=7m/s then increases with ta' which shows that skin friction reduction decreases where the air sheet breaks up. We see similar tendency at xa =5.8m. 1.0 ~ ~ 1 0.8 - 03 .. ~ o 0.6 0.4 0.2 0.0 ~ Xa=0 5m (S1 & s4) is Xa= 1 em (S2 & s5) c Xa=58m (S3 & s6) - - Curve fit (Xa=0 . 5m) - ~Curve fit (Xa=1 .8m) - - - Curve fit (Xa=5.8m) 1.2 1.0 0.8 0.6 cat 0.4 0.2 0.0 1.2 ~ Xa=0 . 5m (S 1 & s4) `` Xa=1.8m (S2 & s5) 5 Xa=5 . em (S3 & s6) - -Curve fit (Xa=0.5m) - - - Curve fit (Xa=1. em) Curve fit (Xa=5.8m) 0.4 1.0 Q8 'I 06 cat Q2 DO ~ "Xa=5.8m (V=5m/s) F ;.xa=~5.8m.(v=7'm/s) , 0.00 1.00 2.00, 3.00 t_a tom) Figure 20 Correction of skin friction measured with (C)xa=s.8m sensors of lOgf full scale. (3) Cf /Cfo The measured skin friction reduction values thus corrected are shown in Fig.21. At xa =O.Sm, the skin friction reduces to zero, because the point was covered with a large air sheet, especially at V=5m/s. At xa =1.8m, the Cf /Cfo value first decreases and 10 l ~ Xa=0 . 5m (V=5m/s) · Xa=0 . 5m (V=7m/s) · ~ · A _ 0.00 1.00 2.00 3.00 t_a (mm) (a) xa =O.Sm : A' ~ L\Xa= 1 . em (V=5m/s) AXa=1.8m (V=7m/s) . · . . . . . l 4.00 0.00 1.00 2.00 3.00 4.00 t_a (mm) (b) xa =1.8m ~ 1 ¢-1 ~ :" Figure 21 Skin friction reduction at 12m-long flat plate ship of NMRI (PP, corrected). (4)Comparisonof Cf /Cfo with Watanabe's results Fig.22 shows the present data compared with those by Watanabe et al. 1998 shown in Fig.17, at V=7m/s. The both data were obtained using PP and at the same streamwise locations. The two results agree well.
11 12 10 Qua 'A Q6 cat Q4 02 , ~ ~ ~ O ~ Il O ~ Xa= 1 . 8m NMRI PP _ OXa=5 . 8m NMRI PP - QXa= 1 . am IHI PP ~ XX~ ; Am TlIT PP 0.0 ID 2.0 3.0 40 t_a (mm) Figure 22 Comparison of skin friction reduction results by NMRI and Watanabe et al. (PP) 2.6 Experiments using a 50m flat plate ship at NMRI The test was carried out twice, in 1999 and in 2000, with different layout. (l)Tested ship and devices In 1999 the 12m ship was extended to 50m by adding the parallel part. At the air injection point at 3.0m from the bow end, an array-of-holes plate (AMP) was set, in order to obtain data for the full-scale test described later. The skin friction sensors were placed at x=3.5m, 4.8m, 8.8m, 16.8m, 24.8m, 28.8m, 32.8m, and 46.8m, where x denotes distance from the bow end. Both 2gf and 10gf sensors were used. The data by the 10gf sensors were corrected in the way described in the previous section. P1 tC Arms ply t4- 8rn) p3 ($~) \l ! HE _m . (A' r ~ eject `on :at the A . (a)Bow half palm Dime pa (~.~3 P(i~30~3~, ' W. ~ , I ~ I I~ ~ I ~1~ I I I I~1 i i ~ I ~ I , ~ I :N Or tnj`cltio'~ ~~ Ted, Ale (b)S tern half Figure 23 50m-long flat plate ship of NMRI (AMP) In 2000 another AHP was set at x=3 1 m, in order to investigate the influence of boundary layer thickness on skin friction reduction effect. This time the sensors were placed at x=3.5m, 4.8m, 8.8m, 31.5m, 32.8m, and 36.8m, i.e., three sensors were placed at the same rela- tive distance downstream of each injection plate. Only 2gf sensors were used. Fig.23 shows the layout of the 50m flat plate ship used in the 2000 test. (2' Cf /Cfo: Ha distribution The streamwise distribution of the measured skin friction reduction at two speeds is shown in Fig.24. The data of the two tests are shown with the same symbols, because they agreed well with each other. The skin friction reduction is greater at V=Sm/s than V=7m/s. At V=Sm/s the reduction persists to the downstream end, while, at V=7m/s, the reduction is almost comparable with that of V=Sm/s immediately downstream of the injection point, but reduces more quickly downstream. I.2 I.0 0.8 to t0.6 cat 0.4 0.2 - ~ ' ~Q ~ ~ V= 5m/s · V=7m/s 1 . ~ . ~ . ~ . ~ ~ 0 10 20 30 Xa (m) 40 50 Figure 24 Streamwise distribution of skin friction ~- duction by microbubbles on 50m flat plate ship of NMRI with air injection at x=3.0m (AMP) (3) Cf /Cfo: Comparison of bow and middle injec- tion In case the drag reduction effect of microbubbles in full scale is estimated using the measurements in circulating tunnels or towing tanks, in which the length scale is considerably smaller than the full scale, the choice of the parameter for air injection rate is impor- tant. For example, Merkle and Deutsch 1990 used the average void ratio in the boundary layer as such pa- rameter. In this case, the amount of injected air needed increases with the length scale, and can result that the application of microbubbles to full-scale ships 11
12 is not feasible. To find out how important, or unimportant, the boundary layer thickness is to the skin friction reduc- tion effect, is also useful for designing air injectors for a full-scale ship. Thus, in the experiment of the 50m flat plate ship, AHPs were placed at x=3.0m (Bow) and 31.0m (Middle). The comparison of the measured skin friction reduction for the two injection cases are shown in Fig.25 (V=7m/s). The reduction by the middle injection is slightly greater. That is perhaps because the boundary layer is thicker there and it helps the bubbles to stay near the wall. Although there is slight difference, the magnitude of the reduction is ap- proximately the same, and it may be stated that the re- duction effect is mainly a function of the distance from the injection point and that the influence of the bound- ary layer is small. 0.8 Q6 0.4 1 ~ ~ ~ a ~ ~ ~ 1 ° Xa=0 . 5m (Bow) O Xa=5 . em (Bow) XXa=0.5m (Middle) Xa=5.8m (Middle) X O X 0 1 2 3 4 ta (mm) 5 Figure 25 Skin friction reduction in bow (x=3.0m) and middle (x=31 m) injection cases of the 50m flat plate ship. V=7m/s. (4' Cf / Cfo; comparison of AHP and PP 1.2 1.0 0.e Cal 1, o.e 0.4 0.2 1~.~.~t it: · ~ ~ °t_a=2rnrn AHP NMRI X<_a=4mm AHP NMRI Qt_a=1.9mm PP Watanabe · t a=2.9mm PP Watanabe 0 10 20 30 Xa (m) 40 50 Figure 26 Cf /Cfo; Comparison of AHP (NMRI) and PP (Watanabe). V=7m/s The streamwise distribution of the skin friction reduction by NMRI using AHP is compared with that by Watanabe et al. using PP at V=7rn/s, as shown in Fig.26. It is clearly seen that the reduction of PP is significantly larger than that of AHP. It seems that, although the question on the accuracy of the lOgf sen- sors has not been answered yet, PP reduces skin friction considerably more than AHP, especially at large down- stream distance from the injection point. 3. Local void ratio distribution of microbubbles It is now a generally accepted idea that there must be many bubbles close to the solid wall, in order to have significant skin friction reduction. The consen- sus on the degree of closeness has not been reached, but Merkle et al. 1987 says, based on their observation of bubble behavior in plate-on-top and plate-on-bottom conditions and the skin friction reduction, that if there is no bubble in the range y+ less than 100 no skin friction reduction occurs, for example. Here, we show only one example of the measure- ment carried out at NMRI (Kodama et al. 20001. Fig.27 shows the system we used to measure the local void ratio as a function of the normal distance from the wall, designed after the one used by Guin (Guin et al. 1996~. The mixture of air and water is taken into a tube facing the stream by suction, and the volume of air and water is measured separately in the two charribers. The local void ratio Car is then calculated using eq.~11), which is similar to eq.~3) for average void ra- tio. fV = Qa Qa +Qw Ally >\ acuum meter A'r:~ Cl~mberA ve o Chamber B | Electromag-/ ,~ ~ ~ ~ I Valve B l Vatcaunukm [low ~~` ° 0 0 opt ° O ° no ~ Suction tube Vacuum pump Figure 27 Local void ratio measurement system at NMRI (Kodama et al. 2000) Fig.28 shows the measured local void ratio on the 12m flat plate ship of NMRI with PP, at the speed of V=7m/s and at the air injection rate of ta=0.71mm. 12
13 At xa =0.Sm there is a sharp and high peak in the vi- cinity of the wall, but at xa =5.8m the peak decreases significantly, which corresponds to the decrease in the skin friction reduction effect. It should be noted that this measurement required several runs in the towing tank to accumulate sufficiently the air and water vol- umes. n an - ~ ~ 0.30 t,O 0.20 0.10 0.00 O °Xa=0.5m ; °g AXa=5. 8m ) o ~^ \. · . °~1 0 10 20 30 40 50 60 7 0 80 90 y (mm) AXa=5 . em Figure 28 Local void ratio distribution on the 12m flat plate ship of NMRI with PP at V=7m/s and ta =0.71mm. 4. Skin friction reduction mechanism of micro- bubbles The skin friction reduction mechanism of micro- bubbles is not yet fully understood. Since the high local void ratio close to the wall surface is related to the skin friction reduction, there is no doubt that the so-called "density effect" plays an important role. There are several observations that bubbles in the turbulent boundary layer modify (decrease) the turbu- lence intensity and thus decrease skin friction. For example, Merkle and Deutsch 1990 showed that the turbulence fluctuations at the wall is reduced by the presence of bubbles near the wall, by using a hot film mounted flush with the surface. More recently, Nag aya et al. used PIV/LIF to measure turbulence in the water part of bubbly flows in the channel shown in Fig.7, by selectively obtaining the image of the tracers that emitted fluorescent light whose wavelength was different from the original laser sheet. According to their measurement, at U=5m/s, both u and v velocity fluctuations increased but the Reynolds stress decreased, with increasing void ratio, as shown in Fig.29. Oh n 5 O- -2 -1 0 1 2 up Figure 29 Measured Reynolds stress distribution across the channel height of HSWT shown in Fig. 7 Nagaya et al. 2002. U=Sm/s. OC: average void ratio. If there are mechanisms other than the density ef- fect for the skin friction reduction by microbubbles, there is a chance that one can maximize the reduction by controlling the bubble size. However, Moriguchi and Kato 2002 changed the bubble diameter from 0.5mm to 2mm at the same flow speed, by changing the channel height only at injection point, but the reduction value did not change. The existence of bubbles makes both measurement and computation difficult. Kawamura and Kodama 2001 carried out DNS of bubbly turbulent channel flow at Rer=360, where the bubble motion and deforma- tion were expressed using VOF, but they got only skin friction increase. Further study is need to clarify then mechanism. 5. Frictional drag reduction by microbubbles (l)Total drag The measured total drag of the 12m flat plate ship in the non-bubble condition is shown in Fig.30. Using Prohaska's method with n=4 and using the data below Fn =O.15, the form factor has been determined to be 1+ K=1.1678. The measured total drag of the 50m flat plate ship in the non-bubble condition is shown in Fig.31. The two tests carried out in 1999 and 2000 gave almost the same result. At very low speed the data in 1999 show laminar behavior. 13
14 0.0045 0.0040 ~ 0.0035 Cal ~ 0.0030 · Ct CfO Schoenherr I\ (l+K)CfO (1+K=1.1678) \~e \ . qb (12) °° b Air injection plate ash 0.0 0.3 0.4 En 0.5 0.6 Figure 30 Total drag of 12m flat plate ship of NMRI in non-bubble condition. 0.0035 0.0030 0.0025 4~ Cal 0.0020 \~ · Ct 2000 (1+K=1.143) ° Ct 1999 (1+K=1.146) C fO (SchoellheIt) (1+K) CfO (1+K=1 . 146) 0°~° ·e o. 0.0015 . ._ 0.00 O. 10 0.20 0.30 0.40 Fn=V/sqrt (gL) Figure 31 Total drag of 50m flat plate ship of NMRI in non-bubble condition. (2)Reduction of frictional drag by microbubbles The reduction of total drag by microbubbles was measured and the reduction was attributed to the reduc- tion of the frictional drag component Rf . Rfo, the frictional drag of the area Sb (see Fig.32), which is downstream of the air injection plate, in the non-bubble condition was estimated using Schoenherr's formula (29~. The result is shown in Fig.33 for the 12m flat plate ship of NMRI (PP), and in Fig.34 for the 50m ship of NMRI (AHP). The horizontal axis shows qb, the injected air rate nondimensionalized by ship's speed and Sb. It is seen that the reduction is smaller at higher speed and that the reduction per unit qb is much higher with the 50m ship than with the 12m ship. The reason for the latter may be that the skin friction reduction effect of microbubble persists for a long dis- tance downstream. 14 07 Ship's hull bottom Figure 32 Sb : Area to be covered with bubbles 1.0 0.9 0.8 0.7 Q6 no 0 ~ ~ 0 ~ o o 0 V=5m/s ~ V=7m/s ° 0.00E+00 1. OOE-04 2. OOE-04 3. OOE-04 4. OOE-04 qb Figure 33 Frictional drag reduction of 12m flat plate ship (NMRL PP) as Q8 ~0.7 0.6 0.5 O ~ to ^~^ ~ ~ A&6 - 0. °V=5m/s 1999 °O ·V=5m/s 2000 aV=7m/s 1999 ~&V=7m/s 2000 O . OE+OO 5 . OE-05 qb 1 . OE-04 1 . 5E-04 Figure 34 Frictional drag reduction of SOm flat plate ship (NMRI, AHP, bow injection) 6. Applicability of microbubbles to ships This chapter aims at discussing the applicability of microbubbles to ships. First, an equations for the net power saving of a full-scale ship is derived by consid- ering the skin friction induction by microbubbles and the power needed to inject them. Second, issues on
15 the practical application of microbubbles to ships are discussed, using the parameters showing in the equation derived. Finally, the net power saving value is esti- mated using available experimental results and typical specifications of a full-scale ship. ( 1 ) Equation for net power saving We assume that a ship runs at speed U with power WO, "O" denoting the non-bubble condition, and by Ejecting bubbles the ship runs at the same speed where p: water density, U with different, hopefully reduced, power W . WO and W are expressed as WO = DoU, W = DU, (13) where Do, D: ship's drags. ~/\D) and ~/\W), the drag and power reductions, are related as (-AW)= (-AD)U, (14) where Do D = DFbO DFb - In order to estimate the net drag reduction effect, one should subtract the power needed for bubble injec- tion from the power gain due to drag reduction. The pumping power Wpump is expressed as Wpump (PgZ + p)Q (20) g: gravity acceleration, z: water depth at injection point, p: dynamic pressure at injection point, Q: volumetric air injection rate. By introducing nondimensional parameters CQ _ Q : volumetric air injection coefficient, USb rz _: water depth ratio ~ L': ship length) AW_W-WO, /\D-D-DO (15) (21) (22) FnU / it;: Froude number, (23) C _ P / I PU2 (24) Let DF be the frictional drag and define P 2 rF - DF /D. (16) Let DFb be the frictional drag of the surface covered with bubbles. The ratio DFb/DF should be ~p- proximately equal to rs-Sb/S (17) where S: wetted surface area of the ship, Sb: surface area covered with bubbles, but by selecting the Sb location near the bow we can expect, using a parameter mb > 1, DFb ID = rFrSmb- (18) Since the drag reduction occurs only in DFb, 15 Wpump of eq.~20) is expressed as Wpump = 2 pU35bCQ~ Z2 +CP), (25) where p: water density. Finally the net power saving ratio rnpS is ex- pressed as, using eqs.(l3) through (25), (-/\W) - Wpump `26) nps W = rS [rF mb (1~~ 2 + C p )] ' DFbO Do n (27) where DO has been nondimensionalizedas CDO - DO /PU25 . (28)
16 For example, rnpS = 0.05 means 5% net power saving. It is the purpose of microbubble studies to maximize this parameter. (2)Issues on the application of microbubbles to ships This chapter discusses the issues on the applicability of microbubbles to full-scale ships mainly by discussing each parameter in eq.~27~. a) lDFb / DFbO This parameter represents the reduction of the fric- tional drag by microbubbles and therefore is the most important. It is not the local value but the integrated value over the entire surface covered with bubbles. Fig.34 shows the values for the 50m flat plate ship in the bow injection case. The frictional drag in that area to be covered with bubbles in the non-bubble con- dition was estimated using the Schoenherr experimental formula shown below. logy ReCFS h I) = 0~242 log ~ (29) The figure again shows that the skin friction reduction is smaller at higher speed. This fact suggests that m- crobubbles should be applied to slow-speed ships. The maximum reduction at V=7m/sec reaches over 20%. The corresponding reduction n the total drag was9%. The reason for the higher decay of the skin friction reduction effect at the higher speed is suspected to be the higher diffusion of the bubbles away from the wall by the higher turbulence intensity and/or smaller bubble size. byrs We should efficiently cover a wide surface area with bubbles. e) rz / Fn2 This ratio, being equal to go / U 2, means that the pumping power Educes quadratically with speed. So the higher the speed, the better the performance. Note that the skin friction reduction effect decreases with speed at the same time. It would also be possible to reduce rz by utilizing the downward flow near the bow. f) Cp By choosing the location for bubble injection or designing the local shape, one can reduce Cp and thus improve the net power gain. g) Hull form As Fig. 1 suggests, the hull form of a large tanker is regarded suitable for microbubbles, because its flat and very wide bottom will help the bubbles injected at the bow stay close to the bottom by buoyancy. It is also important to consider effects of non-horizontal surface, pressure gradient, and surface curvature. h) Sea water Bubbles generated in sea water are in general smaller than those in fresh water with which almost all the laboratory experiments were carried out. This will affect the trajectories and the skin friction reduction effect. i) Propeller performance in bubbly flow As Fig.1 shows, there is a good chance that the bubbles go into a propeller operating at the stern end. Ichikawa and Matsumoto 1995, measured lift and drag of a 2D NNCA4412 wing section in various average void ratio (Fig.35), and found that lift decreased and 1 B |.6 I.4 '.2 ~ 1 0 8 0 6 o.` 0.2. n ~ .- l _ OX . _ ~ 2X _ _ ~ 51 .... .... .... C) rF The wave-making drag, the other major drag com- ponent, increases quadratically with speed, and there- fore the smaller the speed, the greater rF becomes, and the better-suited to microbubbles. With a dis- placement-type ship such as a large tanker at the cruis- ing speed of 14knots (7m/sec), rF reaches 0.8 ~ °,,,,~,; r.~ - C(C - .) 5 2) d) CQ /CDO We should minimize the air volume for a given skin friction reduction. 16 _.w 0.3 0.2-g 0.2 0.15 0.1 0.05 O 0 ~ 10 15 . - ~e of Sty (I ] (a)Lift (b)Drag Figure 35 Lift and drag of NACA4412 wing section in bubbly flow (Ichikawa 1995~.
drag increased by the presence of bubbles. Therefore we should expect that the propeller performance suffers when bubbles go into the propeller, and we should avoid it. The influence of inflowing bubbles on propeller vibra- tion should also be investigated. j) CFD prediction methods In order to predict the drag reduction performance of bubbles at full scale, one must first simulate the flow around a full-scale ship, which is already very difficult. Second, one must predict bubble trajectories by consid- ering the bubbles as a group, which means the two-way coupling approach. (3)Estimation of rnpS rnpS is now estimated. Observing the plots in Fig.34 are almost linear at 7m/s, let us define the slope a b (1 DFb / DFbO )/ CQ CF, the skin friction of a ship, is traditionally related to CFS h ' the skin friction of a flat plate with the same length and area, as CF = (1 + K)CFSCh, (31) where 1 + K iS called the form factor. Using the two equations and eq.~16), rnpS of eq.~27) becomes rnpS = rS rF CQ [mbab - (} K TIC ~ F. 2 + Cp )] Using the parameter values shown in Table 1, in which dimensions are of a large tanker, the bubble area is the front half of the hull surface, mb is given as the ratio of CFS h of 50m and 100m, rF and 1 + K are typical values of a large tanker. Using the values it results that rnpS = 0.046, (33) which means 5% net power saving. If z can be E- duced to Sm somehow, it becomes that rnpS =0.069. Table 1 Parameters for estimating rnpS L (m) B (m) Z (m) U (mls) rS 100 20 7 7 _ 0.5 mb 1 1.093 1 0.8 Cp T ° 1+K 1 1.3 ab 2373 Figure 36 Training ship "SEIUN-MARU", Institute for Sea Training. 5,884 GT, Loa=1 16m, B=17.9m, 10,500PS, l9.5kt. 7. Full-scale microbubble experiment In September 2001, a full-scale microbubble ~- periment was carried out using a training ship "SEIUN-MARU" of the Institute for Sea Training (Fig.36~. It was carried out by a team of researchers who worked under the SR239 "Study on Frictional Drag Reduction Methods for Ships" Esearch project organized by the Shipbuilding Research Association of Japan. The following descriptions are from the SR239 Research Report 2002. The detail of the microbubble study in the project will be published as Kodama et al. 2002 and Nagamatsu et al. 2002. Air, which was supplied by six compressors on deck, was injected through three separate horizontal ducts welded on the bow side, as shown in Fig.37. Each duct has holes of AHP type. The tests were carried out by switching on and off the air injection while keeping the propeller rps con- stant, and the change in the ship speed was converted into power change at constant speed. Fig.38 shows the speed-power curve in bubble and non-bubble condi- tions. Air was injected at full power (max) or half power (1/2 max) from all the three locations (All), and in all the bubble injection cases, contrary to the expec- tation, the engine power increased at constant speed.
18 Figure 37 The bubble injectors and a TV camera BI 1, 2, 3: Bubble injectors No. 1, 2, 3 C: Underwater TV camera From the measurement of thrust and torque in the bubble conditions, it was suspected that due to the bub- ble entrainment, propeller performance was deterio- rated. Therefore, another series of tests were carried out, in which air was injected from a single injector, one by one. And when air was injected only from BIT, the propeller performance was hardly deteriorated, and 3% power saving was obtained, as shown in Fig.39. In this condition, since BI1 from which air was injected was only lm below the water surface, the 2% net power 5000 4000 3000 3 o ~ 2000 i= Cot 1000 O _ o No bubbles ~ x A llmax i - a--- AH 1/2max him ./ ~ ~ . . . . . . . . . 9 10 1 1 12 13 14 15 16 17 18 19 V(kt) Figure 38 Speed-power curve in bubble and non-bubble conditions 18 ~ without bubbles , I · with bubbles 5 ,= 1 7 0 0 o ED ~ 1600 i= 500 400 1~. -. kW} kW . A. 4;7'I ;;4 . V3 curve . , 1.. ~ . . 'i . . s 3.6 13.8 14.0 14.2 14.4 14.6 Ship speed V (kt) Figure 39 3% power saving at about 14 knots with max. air injection from BIT. saving was obtained, by taking into account the hydro- static pressure at BIT. The reason that the bubbles generated at such a shallow location reduced the drag efficiently is that there was downward flow that carried the bubbles down to the hull bottom. This is a good news because the power needed to inject bubbles can be reduced significantly in that way. 8. Conclusions Microbubbles can be called "a big child -- imma- ture, but with bright future". They reduce skin friction significantly both in small scales and in large scales, and the SEIUN-MARU experiment has proved that they can actually reduce drag of a full scale ship with length over lOOm. But at the same time many prob- lems need to be solved before they can be put to practi- cal use. The mechanism for skin friction reduction by microbubbles is not yet fully understood. This causes uncertainty in the way the experiment is carried out and leaves many related problems, such as scale effects, unsolved. To date no numerical simulation has suc- cessfully simulated the mechanism and the skin friction reduction effect of microbubbles. This is a serious disadvantage, since numerical simulation can give us detailed information on the flow, which experiment cannot. Methods that can measure the interaction of bubbles with the turbulence in the boundary layer need to be developed. Simulation methods that can sim~- late the skin friction reduction effect and elucidate the bubble interaction mechanism need to be developed,
too. The application of Microbubbles to full-scale ships poses another series of problems to be solved, such as scale effects, design of bubble injectors, deterioration of propeller performance due to inflowing bubbles, and prediction methods for bubble trajectories along a ship hull. The information on the bubble behavior at full scale is perhaps the most lacking and the most needed. It was a good news that drag reduction by Microbubbles was confirmed in the full scale test using SEIUN-MARU. Further full scale test should be conducted, especially using a ship that has a wide and flat bottom, where bubbles are expected to perform well. Acknowledgements A part of the present study was carried out as the SR239 project organized by the Shipbuilding Research Association of Japan, funded by the Nippon Foundation. The full scale experiment using SEIUN-MARU was carried out solely as the SR239 project. The authors thank the members of the microbubble working group of the SR239 project, especially Prof H. Kato from Toyo Univ., Profs. T. Suzuki and Y. Toda from Osaka Univ., Prof. T. Nagamatsu from Kagoshima Univ., Mr. S. Yamatani from the Institute for Sea Training, Mr. T. Lanai from the Shipbuilding Research Center, Mr. S. Ishikawa from Mitsubishi Heavy Industries, Dr. M. Takai from Sumtomo Heavy Industries, Dr. H. Kamiirisa from Mitsui Engineering & Shipbuilding Co., Ltd., Dr. S. Ogiwara from IHI Marine United, and Y. Yoshida from IHI Aerospace for providing the full-scale data and for valuable discussions. The au- thors also thank Mr. Yabuki, Captain of the SEIUN-MARU, and Mr. K. Yamashita from Chugoku Paint, for the cooperation in the full-scale experiment. References Fukuda, K. et al., "Frictional Drag Reduction with Air Lubricant over Super Water Repellent Surface (2nd report>- Resistance Tests of Tanker and High Length-to-beam-ratio Ship Models --", J. of Soc. of Naval Architects of Japan, vol.186, 1999, pp.7~81. Guin, M.M. et al., "Reduction of Skin Friction by Mi- crobubbles and its Relation with Near-Wall Bubble Concentration in a Channel", J. of Marine Science and Technolo~v, vol. 1, No.5, 1996, pp.241-254. Kato, H. et al., "Frictional Drag Reduction by injecting 19 , , 19 Bubbly Water into Turbulent Boundary Layer", Cavita- tion and Gas Liquid Flow in Fluid Machinery and De- vices, FED -vol. 190,ASME, 1994, pp 185- 194. Kato, H. et al., "Effect of Microbubble Cluster on Tur- bulent Flow Structure" IUTAM Symposium on Me- chanics of Passive and Active Flow Control, Gottingen, 1998. Kawakita, C. and Takano, S., "Microbubble Skin Fric- tion Reduction under the Influence of Pressure Gradi- ents and Curved Surface", J. of the Society of Naval Architects of Japan, vol. 1 88, 2000, pp. 1 1- 21 (in Japa- nese). Kawamura, T. and Kodama, Y., "Direct numerical simulation of interactions between bubbles and wall-turbulence", TSFP-2, Ad Int. Svmp. on Turbu- lence and Shear Flow Phenomena, Stockholm, 2001. Kodama, Y. et al., "Experimental study on microbub- bles and their applicability to ships for skin friction reduction.", Int. J. of Heat and Fluid Flow vol.21, 2000. Kodama, Y. et al., " A Full-scale Experiment on Micm- bubbles for Skin Friction Reduction Using SEIUN MARU Part 1: The Preparatory Study", to be pub- lished in J. of the Society of Naval Architects of Japan, vol. 192, 2002 (in Japanese). Madavan, N.K. et al.," Reduction of Turbulent Skin Friction by Microbubbles", Physics of Fluids, Nb1.27, No.2, pp.35~363, 1984. McCormick, M.E. and Bhattacharyya, R., "Drag Pe- duction of a Submersible Hull by Electrolysis", Naval Engineers Journal, Vol.85, No.2, 1973, pp. 11-16. Merkle, C. L., Deutsch, S., Pal, S., Cimbala, J., and Seelig, W., "Microbubble Drag Reduction", 16th SYm- posium on Naval Hvdrodvnamics, 1987, pp.199-215. Merkle, C. L and Deutsch, S., "Drag Reduction in Liq- uid Boundary Layers by Gas Iniection", Progress in Astronautics and Aeronautics vol.123, 1990, AIAA, pp.35 1-412. Moriguchi, Y. and Kato, H.: Influence of Microbubble diameter and bubble distribution on frictional resistance reduction by microbubbles, J. of Marine Science and Technology vol.7, No.2, 2002 (to be published). Nagamatsu, T. et al., " A Full-scale Experiment on Microbubbles for Skin Friction Reduction Using SEIUN MARU Part 2: The Full-scale Experiment Prepara", to be published in J. of the Society of Naval
20 to be published in J. of the Society of Naval Architects of Japan, vol.192, 2002 (in Japanese). Nagaya, S., Hishida, K., Kakugawa, A., and Kodama, Y., "PIV/LIF Measurement of Wall Turbulence Modifi- cation by Microbubbles", Proceedings of the 3rd SY m- posium on Smart Control of Turbulence, March 2002, Tokyo. Schlichting, H. ,"Boundary-Layer Theory", 6th edi- tion, McGrawhill, 1968. SR239 Research Report, "Study on Frictional Drag Reduction Methods for Ships", Shipbuilding Research Association of Japan, March 2002. Takahashi, T. et al., "Streamwise Distribution of the Skin Friction Reduction by Microbubbles", J. of the Society of Naval Architects of Japan, vol.182, 1997, pp. 1-8 (in Japanese). Takahashi, T. et al., "Experimental study on drag reduc- tion by Microbubbles using a 50m-long flat plate ship" TSFP-2~ 2nd Int. Symp. on Turbulence and Shear Flow Phenomena, Vol.1, June 2001, Stockholm, pp. 175-180. Yoshida, Y. et al., "Distribution of void fraction in bub- bly flow through a horizontal channel: Bubbly bound- ary layer flow, 2nd report", J. of Marine Science Tech- nology 3, 1998, pp.30-36. Watanabe, O. et al., "Measurements of Drag Reduction by Microbubbles Using Very Long Ship Models", J. of Soc. Naval Architects Japan, vol.183, 1998, pp.53-63 (in Japanese). 20
DISCUSSION Prof. C. M. Lee Pohang University of Science and Technology, Korea I suppose the size of the bubbles should have some effect on the reduction in the friction drag. I wonder if you have investigated the effect of the bubble size. AUTHORS' REPLY I understand that the discussion on the effect of bubble size on the skin friction reduction effect has not been settled yet. The main reason for that is that techniques for changing bubble size without changing the flow property such as flow speed are very rare. But recently Moriguchi, Y. and Kato, H. 2002 developed a technique and changed the bubble diameter from 0.5mm to 2mm, but the skin friction reduction effect did not change. Kawamura, T. et al. 2002 developed another technique which generates bubbles of 20 to 40 micron meters diameter, and their preliminary result shows that the reduction effect is significantly greater than that by the bubbles of mm size. So further research is needed to clarify the effect of bubble size on the reduction effect. Kawamura, T., Kakugawa, A., Kodama, Y., Moriguchi, Y. and Kato, H.: Controlling the Size of Microbubbles for Drag Reduction, Proceedings of the 3rd International Symposium on Smart Control of Turbulence, March 2002, Tokyo, Japan, pp. 121-128, (http://www.turbulence- control.grjp/symposium/FY200 1 /index.htm 1) Moriguchi, Y. and Kato, H.: Influence of microbubble diameter and bubble distribution on frictional resistance reduction by microbubbles. J. of Marine Science and Technology (to be published). DISCUSSION Hoyte C. Raven MARIN, The Netherlands The authors are commended with this interesting paper and important work. As mentioned in Section 4, microbubbles seem to reduce the turbulence intensity. Are there experimental data on the effect on the boundary layer velocity profile? microbubble injection susceptibility to flow separation? AUTHORS' REPLY Could it happen that would increase the . ~ Nagaya et al. 2002 measured velocity distributions in the channel flow with bubbles using the PIV/LIF technique, and showed that the Reynolds stress decreases and the mean streamwise velocity distribution approaches laminar profile with increased air injection. Since air is injected normal to the wall, there is possibility, in principle, that it increases susceptibility to flow separation, especially at high injection rate. But the authors think that there was no flow separation in their tested cases, because of the observed strong persistence of the skin friction reduction effect in the downstream direction. Nagaya, S., Hishida, K., Kakugawa, A., and Kodama, Y., "PIV/LIF Measurement of Wall Turbulence Modifi-cation by Microbubbles", Proceedings of the 3rd Sym-posium on Smart Control of Turbulence, March 2002, Tokyo. (http ://www.turbulence- control.gr jp/symposium/FY200 1 /index.htm 1) 7 No.2, 2002
