experimental study on the wake structure behind a ... · hideo oshikawa and toshimitsu komatsu /...

Procedia - Social and Behavioral Sciences 218 ( 2016 ) 161 – 170

Available online at www.sciencedirect.com

ScienceDirect

1877-0428 © 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).Peer-review under responsibility of Dept of Transportation Engineering, University of Seoul.doi: 10.1016/j.sbspro.2016.04.019

11th International Conference of The International Institute for Infrastructure Resilience and Reconstruction (I3R2)

: Complex Disasters and Disaster Risk Management

Experimental Study on the Wake Structure behind a Cylindrical Rotator with Asymmetric Protrusions in a Unidirectional Flow

Hideo Oshikawaª , Toshimitsu Komatsub

ªDepartment of Civil Engineering and Architecture, Saga University, 1 Honjo-machi, Saga 840-8502, Japan b

Department of Urban and Environmental Engineering, Kyushu University, 744 Motooka, Nishi-ku, Fukuoka 819-0395, Japan

Abstract Wake structures behind a cylindrical rotator with asymmetric protrusions were investigated through a laboratory experiment. Such a cylindrical rotator could prevent a flood disaster due to the accumulation of driftwood at bridges. In this study, a rotator composed of a cylinder with five quarter-cylindrical protrusions was placed in a unidirectional flow. The rotator could rotate freely under hydrodynamic forces. The results indicate that vortices shed from the rotator affected flow structures behind the rotator. The transverse distribution of the time-average longitudinal velocity of the flow with respect to the rotator was asymmetric about the longitudinal centerline. In the accelerating area with rotation of the rotator, five vortices were obviously shed from each quarter-cylindrical protrusion in each period. Where the flow and rotation were in opposite directions, vortices formation was disrupted. A longitudinal series of vortices merged downstream, ultimately becoming a single vortex. Therefore, the wake flow becomes asymmetric in an area behind the rotator where a significant torque acts. © 2015 Hideo Oshikawa. Published by Elsevier Ltd. Selection and/or peer-review under organizing committee of I3R2 2015 Keywords: Cylindrical rotator; wake; vortex; Reynolds stress; driftwood 1. Introduction

In recent years, the frequency of torrential rains resulting in slope failure, bank erosion, and other problems has been increasing. In particular, there are many cases where driftwood flowing in medium and small rivers builds up against bridges that span the river, damming the river because the span length of the bridge is too small in some cases. As a result, overflow occurs that inundates nearby houses, shore areas, and other places resulting in massive damage, even when the flow discharge does not exceed the design high-water discharge.

Research into driftwood and measures for dealing with it have therefore become a pressing issue [1]. However, although there are examples of research into the accumulation of driftwood at bridges and houses [2-7], there are few examples of investigation into specific measures for reducing the damage caused by driftwood [1]. A detachable parapet of a bridge was proposed as a countermeasure to reduce the accumulation of driftwood and refuse during flooding [8].

Against this background, we previously proposed a method for preventing the buildup of driftwood and refuse at bridges by installing a rotating cylinder called the "rotator," which is an asymmetric structure set in front of the bridge piers, and a similar method for preventing the buildup of driftwood and refuse at bridges by wrapping the bridge piers in a rotating caterpillar fitted with an asymmetric structure [9]. These methods use the massive energy of the water flow during a flood to turn the flow direction of driftwood around a bridge pier by the rotator or the caterpillar and to promote the washing out of driftwood that would otherwise build up against the bridge pier. If there is no torque around a bridge pier without the rotator or the caterpillar, driftwood and refuse are likely to balance and accumulate there. Corresponding author. Tel.: +81-952-28-8685 ; fax: +81-952-28-8699 .

E-mail address: [email protected]

© 2016 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).Peer-review under responsibility of Dept of Transportation Engineering, University of Seoul.

http://crossmark.crossref.org/dialog/?doi=10.1016/j.sbspro.2016.04.019&domain=pdf

http://crossmark.crossref.org/dialog/?doi=10.1016/j.sbspro.2016.04.019&domain=pdf

162 Hideo Oshikawa and Toshimitsu Komatsu / Procedia - Social and Behavioral Sciences 218 ( 2016 ) 161 – 170

In that regetting cain front ocylinder wasymmetrputting th

Accowith the driftwoodpier wrap

Reseaxis windturbine inknowledgof the rotfluid intro

Furtincreasingway, by awind turbmore effispiral cylby Ito et with the a 2. Metho

A lin

in the expdepth at amean flowFroude nudiameter.where g i

Table 1. E

The

protrusionspacing (plate instchannel wFig. 1). Tdependincylindrica

esearch [9], waught on the bof a bridge mowas installed. ric shape, andhis technologyordingly, in thaim of under

d at bridge piepped in a rotatearch related d turbines sucn particular isge gained fromtator examineoduced into anthermore, in gly come to bapplying electbine rotate froiciently, usinglinder blades cal. [17] on theaforementione

ods

near open chaperiments. Tha point A (refew velocity at tumber. Re is d. Note that D fis the gravitati

Experimental coQ (m3/hr) 75.0

rotator is a cns formed by (refer to Fig. 2talled spanninwith longitudinThe cylindricang on the curral protrusions

we investigatbridge by flowodel and comHowever, the

d there are aly into practicahis study we rstanding and ers. Note thatting caterpillarto this work,

ch as those ofs relatively sim these previoed in this resend emitted frorecent years

be used in pratric power to rom the aerodyg natural energcorrespondinge technology ed Savonius w

annel of lengthhe experimenter to Fig. 1, 50the vertical crdefined as UDfor the rotatorional accelera

F

onditions h (c 12

cylinder of didividing a cy

2), and the sung the water cnally 950 cm

al structures usrent and, for attached.

ed the effectiwing driftwoodmparing the re

ere is little reslmost no exam

al use. first experimeimproving th

t this investigr of the type mwhich treats

f the Darrieusimilar to the sous studies wiarch is greatly

om the interior, spiral Magactice [15-18]rotate cylindeynamic lift prgy alone, by ug to the fins infor rotating sp

wind turbines.

h 1400.0 cm, wtal conditions 0.0 cm upstreross-section coD/ , where isr is the diamettion.

Fig. 1. Side view

cm) 2.0

iameter D = ylinder of radiupport post of channel. A cydownstream fsed in the expcomparison,

iveness of thd models in ansults with thosearch that tremples of rese

entally investhe effectiveneation is also a

mentioned earlcylinders rota

s and Savoniushape of the ill serve as a uy different thr of the cylind

gnus wind tu]. These winders that have sroduced. Howusing the rotan regular winpiral Magnus

width 60.0 cmare shown in

eam of the lateontaining the s the kinematiter of the cylin

w of the experim

U (cm/s) 29

5.0 cm and lius k = 1.0 cmf the rotator, wylindrical strufrom a honeycperiments wera non-rotatin

he rotating bon open water cose from otheats water flow

earch that can

igated the basss of installinapplicable to lier [9]. ating due to f

us types [10-1rotator examuseful referen

han that of theder in the windurbines, whichd turbines genpiral-shaped f

wever, it may ator from this nd turbines. A

wind turbines

m and height 6n Table 1, wheeral center of tpoint A, Re is

ic viscosity condrical part. F

mental apparatu

1.4

length 12.0 cmm into four equwhich is an axucture was inscomb to contrre a rotator thang cylinder of

odies for prevchannel wherer cases, such

w around a cyln serve as ref

sic wake strucng rotators to the flow arou

flow, includes4]. The shape

mined in this rnce in some we wind turbined turbine. h utilize the

nerate electricifins attached, be possible tresearch insten experimentas without mot

60.0 cm, as shoere Q is the flthe cylinder), s the Reynoldoefficient of wFr is defined as

us

Re 4 X 104

m, to which ual parts havexle of rotationstalled laterallol water surfaat is not contrf the same siz

venting driftwe a rotator waas when a nolindrical rotatferences for t

cture of rotatiprevent the b

und a cylindri

s research intoe of the Savoresearch. Alth

ways, the wakee because the

e Magnus effity in a highlywhich makes

to generate poead of the impal study was tors by combi

own in Fig. 1,flow rate, h isU is the cross

ds number, andwater and D thas U/(gh)1/2 for

Fr 0.27

five quarter-ce been attachen, is fixed by ly in the midace disturbancrolled and rotaze without th

wood from as installed on-rotating or with an he aim of

ing bodies buildup of ical bridge

o vertical-onius wind hough the e structure wind is a

fect, have y efficient the entire ower even permeable conducted ning them

, was used the water

s-sectional d Fr is the

he cylinder r a given h

cylindrical ed at equal an acrylic

ddle of the ce (refer to ates freely

he quarter-

Tim

laser senslaser signultrasoniccomponenfrequency

Figuinstantanemeasurem(y=0, ±2.

3. Result

3.1. Mean

Figuvelocity oexperimeforth are the meanwhere y =one direcin the arefollows thregion. Osmaller th

me-series data osor installed unal and was fc flow velocimnts of flow vy was 100 Hz,ure 3 shows teous flow vel

ment lines (x=5, ±5.0, ±7.5,

ts and Discus

n flow velocity

ure 4 shows thof the rotator

ent—except forepresented a

n flow velocity= 0, the flow vction, and the ea near the rohe main flow

On the other hahan in the y >

of the rotatingunder the rotafound to be frmeter was usevelocity were, and the measthe definitionlocity compon=7.5, 10.0, 12±10.0, ±15.0

Fig. 3. Pla

sions

y and turbulen

he distributionand cylinder

or the frequenas dimensionley distribution velocity distribflow velocity

otator (e.g., y/due to the or

and, in the y <0 region.

Fig. 2. Rotator

g phase and floator. The mear =0.84 Hz. Thd for measurin

e measured asurement dura

ns of the O-xynents (us, vs,

2.5, 15.0, 20.0cm) in the y d

ane view of the

nce structure

n across the chjust behind e

ncy spectra (Fess quantities n

for the cylindbution for the u is larger in

/D=±0.5). In tientation of th

< 0 region, sin

r (left: plane vie

ow velocity foan rotational fhe tip speed ing the flow vat a height ofation was 6 myz coordinatews). The mea

0, 30.0, 40.0 direction.

coordinate syst

hannel of the teach cylinder Figs. 6 and 8normalized byder is left–rigrotator is asy

n the positive ythe y > 0 reghe protrusionsnce the rotator

ew; right: side v

for the rotator frequency of tratio was therelocity, and thf 3.5 cm fromin at each poin

e system and asurement poscm) in the x

tem and the me

two horizontalat x/D = 1.5.

8)—the spatiay D and/or U,

ght symmetricmmetric. Thisy region than

gion, since thes, it is understr rotates in the

view)

were synchronthe rotator warefore 2 fhe horizontal dm the channent. the direction

sitions were adirection and

asurement poin

l components In the figure

l coordinates,, respectively.about the cen

s occurs becauin the negativ

e rotator rotattood that u is e direction tha

onized by usinas determinedfr (D/2+k)/U =distribution oel floor. The

ns of each of at the intersecd 11 measurem

nt

(u, v) of the mes for the resu, flow velocit. It is clear thanter line of thuse the rotatorve y region, ptes in the direlarger than in

at opposes the

ng a digital d from the =0.64. An f the three sampling

f the three ctions of 7 ment lines

mean flow ults of this ty, and so at whereas he channel r rotates in articularly ection that n the y < 0 flow, u is


Nexturbulenc

Here, u’,averagingalthough term on trotation, [19]. Howattempted

Simisymmetribecause t

We asymmetrasymmetrconfirmedturbulencmean flowflow occufrom the

Fig. 4. L

t, Fig. 5 showce, Pr, under th

v’, and w’ ag. The spatialPr of the wa

the right-handthe spatial dewever, the firsd performing c

Fig.

ilar to the meic on y = 0, wthe rotator rotaalso investigaric turbulent ric twin peak d. In this figu

ce. Furthermorw velocity beurs in the y dirvalue of v at y

Lateral distribut

ws the distribuhe same condi

21k

Pr

are the fluctul derivatives iake flow for ad side of Eq. erivative of v wst term was arcalculations, a

5. Lateral distri

ean flow velocwhile those fates in a singlated the relatiodistribution idistributions

ure, k at bothre, these pointecomes large rection accomy = 0 being alm

tions of mean f

ution across titions as for F

'' 22 vu

''' uyuvu

ating parts ofn the equatio

any cylindrica(2), in this rewas also consround one ordand the contrib

ibutions of k an

city distributiofor the rotatore direction. on between thin Fig. 5 arowith a maxim

h points becots also match tin Fig. 4. Bec

mpanying rotatmost zero for

flow velocity ju

the channel oFig. 4. In this r

'2w

'xvv

f us, vs, and won for Pr are aal two-dimensesearch considsidered becauder of magnitubution of the s

nd Pr just behind

on in Fig. 4, tr were asymm

he asymmetricound the rotatmum value atmes larger ththe points whcause of thistion and resulthe cylinder a

ust behind the ro

of the turbulenresearch, k and

ws, respectiveapproximated sional structurdering the dis

use it was thouude larger thasecond term to

d the rotator an

the distributiometric, as sho

c mean flow vtor. For botht y/D = 0.5 anhan in the surhere the absolu

left–right asylts in advectivand considerab

otator and cylin

nt flow energyd Pr are expres

)1(

)2(

ely, and the oby a differen

re is often appstortion of theught that v mian the second o Pr was smal

d cylinder (x/D

ons of k and Pwn in Fig. 5.

velocity distribPr and k of

nd a local marroundings duute value of thymmetry of the transport, wbly negative fo

nder (x/D=1.5)

y, k, and prodssed by Eqs. (

overbars reprence method. Iproximated be flow arisingight also haveterm when w

ll.

D=1.5)

Pr for the cyli. Naturally, th

bution in Fig.f the rotator aximum at y/Due to the prodhe spatial gradhe distributionwhich can be ufor the rotator.

duction of (1) and (2).

esent time In general, y the first

g from the e an effect

we actually

inder were his is also

4 and the in Fig. 5,

D = -1 are duction of

dient of the ns, a mean understood


3.2. Perio

Figuy/D = 0.5and v*, tnondimenand the pits synchr

Ther

frequencyrotator hathe rotatoalso chan

Wheas Fig. 6,of the rotfixed cylrotation trotator, re

Vert

amplitudecylinder structure

odic variation

ure 6 shows th5 near the rotathose are nonsionalized by

phase-averagedronized norma

Fig. 7. N

re are five way of 4.2 Hz, was five protrusor rotates freelnges slightly aen the peak fre it was found tator fr = 0.84linder. Howevtime-series daegardless of w

tical flow vee of fluctuatiwas sufficienaround the cy

n characteristi

he frequency sator, and Fig. 7ondimensionaly the period Td horizontal valized flow ve

Fig. 6. F

Normalized pha

aves both u* awhich is 5 timsions, the five ly depending round fr = 0.8equency of thto be 0.97 Hz Hz, the rotatver, the perioata from the lawhether this ki

Fig. 8. Fr

elocity did noon of vertica

ntly small comylinders is not

ics of the flow

spectra Fu(f), F7 shows the tilized by U,

T (=1/fr) of thevelocities wereelocity data.

requency spectr

ase-averaged ho

and v* in Fig. mes the rotatio

waves in Figon the curren4 Hz. e flow velocitz (refer to Figtional frequenodicity investiaser and its syind of vortex s

requency spectr

ot affect the al flow velocimpared with Fdiscussed fur

w

Fv(f), and Fw(fime series of tat the same

e rotation on the calculated b

tra just behind t

orizontal velocit

7, and the sponal frequency. 7 are induce

nt in this resea

ty spectra aroug. 8), and sincency of the rotaigated in thisynchronized fshedding effec

ra just behind th

horizontal wity ws was smFu(f) and Fv(f

rther.

(f) of the flow the phase-aver point with he vertical axi

by using the ro

the rotator (x/D=

ty just behind th

pectral power y (fr = 0.84 Hd by the protr

arch and is no

und the cylinde this is relativator may be a work is basflow velocity,ct exists.

he cylinder (x/D

wake structurmall since Fw(f) in Figs. 6

velocities us, raged dimensidimensionles

is. Time interpotation signal

=1.5, y/D=0.5)

he rotator (x/D=

of each compHz) of the rotarusions on the t controlled, t

der was determvely close to taffected by voed on the ph so it arises f

D=1.5, y/D=0.5)

e of the cyliw(f) around bo

and 8. There

vs, and ws at ionless flow vss time, t/T, polation was pdata from the

=1.5, y/D=0.5)

ponent dominaator in Fig. 6.e rotator. Notethe rotational

mined at the sthe rotational ortex sheddinghase relationshfrom the rotat

)

indrical strucoth the rotatoefore, the ver

x/D = 1.5, velocity u*

which is performed e laser and

ates at the Since the

that since frequency

same point frequency

g from the hip in the tion of the

cture. The or and the rtical flow

Fig. 9. Normalized phase-averaged horizontal velocity component just behind the rotator at x/D=1.5 [i): /U ; ii): <v>/U ]

Fig. 10. Normalized phase-averaged horizontal velocity component behind the rotator at x/D=2.5 [i): /U ; ii): <v>/U ]

Fig. 11. Normalized phase-averaged horizontal velocity component behind the rotator at x/D=8.0 [i): /U ; ii): <v>/U ]

t

y/D

0 0.2 0.4 0.6 0.8 1-3

-2

-1

0

1

2

3u/u: -0.35 -0.2 -0.05 0.1 0.25

B

t/T

y/D A

C DE

/U

i)

1

t

y/D

0 0.2 0.4 0.6 0.8 1-3

-2

-1

0

1

2

3u/u: -0.35 -0.2 -0.05 0.1 0.25

B

t/T

y/D A

C DE

/U

i)

1

B

t/T

y/D A

C DE

/U

i)

1

t

y/D

0 0.2 0.4 0.6 0.8 1-3

-2

-1

0

1

2

3v/u: -0.35 -0.2 -0.05 0.1 0.25

t/T

y/D

<v>/U

b c d ea

ii)

t

y/D

0 0.2 0.4 0.6 0.8 1-3

-2

-1

0

1

2

3v/u: -0.35 -0.2 -0.05 0.1 0.25

t/T

y/D

<v>/U

b c d ea

ii)

t

y/D

0 0.2 0.4 0.6 0.8 1-3

-2

-1

0

1

2

3u/u: -0.25 -0.1 0.05 0.2

t/T

A B C DE

/U

i)

y/D

1

t

y/D

0 0.2 0.4 0.6 0.8 1-3

-2

-1

0

1

2

3u/u: -0.25 -0.1 0.05 0.2

t/T

A B C DE

/U

i)

y/D

1

t

y/D

0 0.2 0.4 0.6 0.8 1-3

-2

-1

0

1

2

3u/u: -0.06 -0.03 0 0.03

Y/D

y/D

t/T

/U

i)

1

t

y/D

0 0.2 0.4 0.6 0.8 1-3

-2

-1

0

1

2

3u/u: -0.06 -0.03 0 0.03

Y/D

y/D

t/T

/U

i)

1

t

y/D

0 0.2 0.4 0.6 0.8 1-3

-2

-1

0

1

2

3v/u: -0.3 -0.15 0 0.15 0.3

c dbae

t/T

y/D

<v>/U

ii)

t

y/D

0 0.2 0.4 0.6 0.8 1-3

-2

-1

0

1

2

3v/u: -0.3 -0.15 0 0.15 0.3

c dbae

t/T

y/D

<v>/U

ii)

t

y/D

0 0.2 0.4 0.6 0.8 1-3

-2

-1

0

1

2

3v/u: -0.1 -0.06 -0.02 0.02 0.06 0.1

y/D

t/T

<v>/U

ii)

t

y/D

0 0.2 0.4 0.6 0.8 1-3

-2

-1

0

1

2

3v/u: -0.1 -0.06 -0.02 0.02 0.06 0.1

y/D

t/T

<v>/U

ii)

For each of the x/D measurement lines, the variation in dimensionless phase-averaged velocity was extracted by subtracting the dimensionless mean values (u/U, v/U) from each of the dimensionless phase-averaged flow velocities (u*, v*). The variation is indicated by <>. The nondimensionalized results as representative measurement lines are shown in Figs. 9 to 11. The closed solid black curves in the figures represent the approximate positions and sizes of the individual vortices, and the arrows indicate the direction of vortex rotation, that is, the direction of flow. Furthermore, the dotted closed curves similarly show examples of vortices where there is a possibility of merging. Note that in each diagram, i) shows /U and ii) shows <v>/U. Although these diagrams show the periodic time lapse of the flow as detected on each of the measurement lines, this corresponds to drawing a plan view of the coherent flow structures, so-called vortices, for the case where the variations detected on the measurement lines are assumed to continue as they are by advection.

In Figs. 7, 9, and 10, symbols are assigned to each of the 5 waves produced by the protrusions for convenience. Uppercase letters are assigned to the waves for and lowercase letters are assigned to the waves for <v>. Then the uppercase and lowercase letters are associated such that, for example, wave A and wave a refer to the same wave, that is a vortex, generated by the same protrusion. If we look at wave B and wave b at x/D = 1.5 near the rotator in Fig. 9, wave b takes a positive peak where the variation in wave B has a phase of zero (t/T 0.27) at y/D = 0.5, which is dominated by the amplitude of the variation. If we consider the general rotation direction of vortices separated from such a cylindrical structure, we can understand that vortices exist as shown by the solid lines in Fig. 9. 3.3. Region where the rotation direction and main flow direction are the same (y > 0)

First, we describe points such as that it can be seen that the features corresponding to the same vortex are different at a glance, as can be seen in the vortices that are significantly flattened in but are not quite so distorted in <v> in Figs. 9 and 10. For example, if we investigate at y/D = 0.5 and 1.0, where five cyclic variations dominate, flattened vortices occur in as a result of the vortices being stretched out in the x direction because the mean flow velocity u at y/D = 1.0 far from the cylinder is faster than u at y/D = 0.5 (refer to Fig. 4). Note that although the flattened vortices in Fig. 9 i) appear to be tilted counterclockwise, the actual flattened vortices are tilted clockwise. For <v>, mean velocity shear is small since the absolute value of mean velocity is small, as can also be seen from Fig. 4, so the distortion of the vortices is small. The shape of the actual vortices is significantly flattened—as can be envisioned by considering the combination of Fig. 9 i) and ii)—and they are also deformed in the y direction, making the shape even further deformed at some point downstream, where the vortices become more like a crescent shape than a flattened ellipse.

We now discuss transformations accompanying downstream flow of the vortices by comparing Fig. 9, where x/D = 1.5, with Fig. 10, where x/D = 2.5. In Fig. 10 ii), the positive peaks of waves a and d become smaller. Therefore, waves e and a and waves c and d continue to progressively form single vortices, as indicated by the dotted lines in the figure. Furthermore, it can also be seen from Fig. 10 i) that the negative peak disappears from wave A, and the positive regions of waves E and A progressively form a single region. It was thus inferred that merging of the vortices occurred.

In Fig. 11 i) at the downstream position x/D = 8.0, a single set of positive and negative regions exists in , for example, at the period of one rotation that can be seen at y/D = 1.0. This also forms the same single set of variations in <v> in Fig. 11 ii). As a result, at a position somewhat downstream, at x/D = 8.0, a single wave corresponding to the period of one rotation was formed as a result of repeated merging of vortices. That is possibly the disappearance of weak vortices.

3.4. Region where the rotation direction and main flow direction are different (y < 0)

In the region of y < 0, where the rotation opposes the main flow, five clear waves organizing vortices were not observed even in Fig. 9 near the rotator, and the amplitudes of the waves were smaller than in the y > 0 region. In other words, the vortices that occur in the region of y < 0 are weaker than the vortices in the region of y > 0. It is inferred that either merging has already begun upstream of x/D = 1.5 or clear vortices do not necessarily occur from all five the protrusions, and there may be cases where the passage of a protrusion is accompanied by only a weak perturbation not a vortex.

In the region of y < 0 slightly downstream in Fig. 10, both and <v> have almost precisely a single variation in the period of one rotation. In other words, in the region of y < 0, the effect of the protrusions disappears even faster than it does in the y > 0 region. 3.5. Overall structure of mean flow and turbulence

From Figs. 9 to 11, periodic velocity variations become significant in the longitudinal vertical cross-section at y/D = 0.5 in the region of y > 0 and in the longitudinal vertical cross-section at y/D = -1.0 in the region of y < 0.

It is thereregion of

Figuphase-avethis resea

Fig.

Figu

vortices pcould alsdownstresides of yAlso in longitudincharacteri

Finabehind thshows thpointwisey = 0, andpointwiseThe pointy/D = -1significan

efore thoughtf y > 0 flow doure 12 shows teraged flow varch, <k> is ex

Fig. 12. Lo

. 13. Longitudin

ure 12 shows pass as in the so be expecteam direction.

y = 0, the vortiFig. 13, the nal measuremistics of the wally, Fig. 14 she rotator on that behind thee symmetric ad so a torque e symmetry ints where the R.0 are consistnt torque acts

t that from theownstream neathe longitudinvelocity, and Fxpressed by Eq

k

ongitudinal tran

nal transformati

asymmetric tdistribution o

ed from the r In Fig. 12, aices become wmean flow v

ment range whake flow separ

shows the lonthe xy plane, e cylinder for about the origidoes not act. In the range ofReynolds stretent with the in the wake fl

e rotation of ar y/D = 0.5 an

nal transformaFig. 13 showsq. (3).

21 2u

sformation of la

ion of lateral di

twin peak disof k consistingesults in Figsalthough the rweak at the povelocity distrihere x/D 8rated from a cygitudinal tranwhich is particomparison.

in, and the forIn contrast inf x/D < 6, corss has a posititurbulent ene

low region of

the rotator, thand those in thation of the sps that of the m

22 v

ateral distributi

istributions of l

stributions thag of all freques. 9 to 11, thregion where osition x/D = 6ibution has b8.0, a velocitcylinder are mansformation oficularly repreThe Reynold

rces are nearlyFig. 14, the R

rresponding toive peak for yergy distributithe rotator.

he vortices oche region of y <anwise distribmean flow ve

2w

ions of normaliz

ongitudinal me

at have local ency componehe local maxim

the effects of6.0 since the pecome nearly

ty defect regiaintained [19].f the spanwissentative of th

ds stress arouy balanced onReynolds streso the mean floy > 0 at y/D =ion. From the

ccurring from< 0 flow down

bution of turbulocity in the m

)3(

zed <k> behind

an flow velocity

maxima at y/ents shown in ma is greatlyf the vortices eaks in <k> h

y symmetric ton remains in. e distribution he effects of tnd the cylind

n both sides ofss around the ow velocity re

= 0.5 and a nege above, it ca

m each protrusnstream near yulent energy <main flow dir

d the rotator

ty behind the ro

/D=0.5, -1.0 Fig. 5. Furthe

y attenuated toremain differ

have nearly disthere. Note thin Fig. 13, th

n of the Reynothe rotator, an

der in Fig. 15f the central arotator does nesults describgative peak fo

an be understo

sion in the y/D = -1.0. <k> on the rection. In

otator

where the ermore, as oward the rs on both sappeared. hat in the hat is, the

olds stress nd Fig. 15 5 is nearly axis where not exhibit ed earlier. or y < 0 at ood that a

4. Conclu

In thdue to flopier by in(1) Vor(2) Vor

flowFurperi

(3) Theaxisqua

In futhe resultaccumulapiers.

Fig. 14. Lo

Fig. 15. Lon

usions

his research, wow as part of nstalling a cylirtices are genertices in the rew (y>0) flow rthermore, theiod correspone effect of thes accompanyiantities are asyuture researchts of this res

ation of driftw

ongitudinal tran

ngitudinal trans

we conducted the developm

indrical rotatoerated by eachegion where thnear y/D = 0

e vortices mernding to the roe vortices exteing the rotatioymmetric in thh, it will be nesearch, and towood by using

nsformation of l

sformation of la

an experimenment of technoor. The followh of the five prhe direction o

0.5, and vorticrge as they fltational period

ends down to on of the rotahis region wheecessary to exao ultimately ug the rotator

lateral distribut

ateral distributio

ntal investigatology for prev

wing results werotrusions as a

of rotation of tces that occurlow downstred of the rotatoaround x/D =

ator, and the ere a significaamine the appunderstand anwhen taking

ions of Reynold

ons of Reynold

tion of the wakventing the acere obtained.a result of rotathe rotator is tr on the oppoeam, eventuallor. = 6.0 with the mean flow ve

ant torque actspropriate instand improve tinto account

ds stress behind

ds stress behind

ke structure occumulation o

ation of the rothe same as th

osite side (y<0ly forming a

asymmetric felocity and ch.

allation positiothe effectiventhe effects o

d the rotator

the cylinder

of a cylinder thof driftwood a

otator. he direction o0) flow near ysingle variati

forces about tharacteristic t

on of a rotatorness for preveof driftwood a

hat rotates at a bridge

f the main y/D =-1.0. ion with a

the central turbulence

r based on enting the and bridge


Acknowledgements

The authors thank Naoki Terao and Kazuo Fujita for their help in performing the experiment. References

[1] Komatsu, T. (editorial supervision), Foundation of River & Watershed Environment Management (2009): Driftwood and Disaster, Gihodo Shuppan Co., Ltd., ISBN978-4-7655-1759-1 C3051, 273p. (in Japanese).

[2] Nakagawa, H., Inoue, K., Ikeguchi, M. and Tsubono T. (1994): Behavior of Drift Wood and Its Dam up Process, Annual Journal of Hydraulic Engineering, JSCE, Vol.38, pp.543-550 (in Japanese).

[3] Matsumoto, K., Kobatake, S., Shimizu, Y., Ishida, K., Konnai, T. and Ioakim, I. (2001): A Study of Accumulation of a Lump of Floating Logs on Bridge Pier, Annual Journal of Hydraulic Engineering, JSCE, Vol.45, pp.925-930 (in Japanese).

[4] Sakano, A. (2003): Hydraulic Study on Accumulation of Drifting Wood at a Bridge During a Flood, Technical Note of National Institute for Land and Infrastructure Management, No.78, 87p. (in Japanese).

[5] Abe, S., Watanabe, Y. and Hasegawa, K. (2005): Flood Damage to Bridges on Saru River from Typhoon Etau (Typhoon No.10), 2003, Advances in River Engineering, Vol.11, pp.109-114 (in Japanese).

[6] Shimizu, Y., Osada K. and Takanashi, T. (2006): Numerical Simulation of the Driftwoods Behavior by Using a DEM-Flow Coupling Model, Annual Journal of Hydraulic Engineering, JSCE, Vol.50, pp.787-792 (in Japanese).

[7] Gotoh, H., Ikari, H., Sakai, T. and Oku, K. (2007): 3D Simulation of Blocking of Bridge in Mountain Stream by Drift Woods, Annual Journal of Hydraulic Engineering, JSCE, Vol.51, pp.835-840 (in Japanese).

[8] Fujimori, Y., Ochi, Y., Hayama, S., Shiraishi, H. and Watanabe, M. (2008): A Drift Wood Hazard and its Countermeasures in Steep Small-scale Rivers, Annual Journal of Hydraulic Engineers, JSCE, Vol.52, pp. 679-684, (in Japanese).

[9] Oshikawa, H. and Komatsu, T. (2011): Experimental Study to Prevent River Stoppages Due to Driftwood, Proceedings of the 34th IAHR World Congress, ISBN 978-0-85825-868-6, pp.307-314.

[10] Tanaka, F., Kawaguchi, K., Sugimoto, S. and Tomioka, M. (2008): Influence of Wind Section and Wind Setting Angle on the Starting Performance of a Darrieus Wind Turbine, Transactions of the Japan Society of Mechanical Engineers, Series B, Vol.74, No.739, pp.624-631, 2008 (in Japanese).

[11] Savonius, S.J. (1931): The S-Rotor and Its-Application, Mechanical Engineering, Vol.53, No.5, pp.333-338. [12] Yamagishi, M. (2004): Phase Averaging Analysis of the Velocity Profiles in the Wake of a Savonius Rotor, Journal of the Visualization

Society of Japan, Vol.24 Suppl. No.1, pp.79-82 (in Japanese). [13] Yamagishi, M. (2005): Visualization of Flow Field in the Wake of a Savonius Rotor by using Single Hot-wire Probe, Journal of the

Visualization Society of Japan, Vol.25 Suppl. No.2, pp.147-150 (in Japanese). [14] Hayashi, T., Li, Yan and Hara, Y. (2005): Wind Tunnel Tests on a Different Phase Three-Stage Savonius Rotor, JSME International

Journal, Series B, Vol.48, No.1, pp.9-16. [15] http://www.mecaro.jp/ [16] Shuchi, S., Murakami, N. and Ito, J. (2005): Development of a wind turbine utilizing magnus effect for rotating cylinders with spiral

structures, Proceedings of Japan Wind Energy Symposium, Vol. 27, pp.177-179 (in Japanese). [17] Ito, A., Nishizawa, Y. and Ushiyama, I. (2007): An Experimental Study of the Savonius Type Magnus Wind Turbine, Proceedings of

12th National Symposium on Power and Energy Systems, The Japan Society of Mechanical Engineers, No.12, pp.125-126 (in Japanese). [18] Nishizawa, Y., Kawashima, T. and Ushiyama, I. (2007): An Experimental Study of the Savonius Type Magnus Wind Turbine, Journal

of Japan Wind Energy Association, Vol.31, No.1, pp.83-87 (in Japanese). [19] Tennekes, H. and Lumley, J.L. (1972): A first course in turbulence, The MIT Press, 300p..

experimental study on the wake structure behind a ... · hideo oshikawa and toshimitsu komatsu /...

Documents