-n -: : u ti;^:: ivl 'tE fi - - t:ty. : PA: -SUDY l @: FORl . AT 'Jfl, *g. .: iXs OT 1ET .. 0 DT DI OF sUR2 GT HE VO-2';RT2CITYAM AD D ZL FRSE SURFAE-C IX - MThCZ A DISFHPANi i2RI IENT by John DI. MhLlhaljan :wK : jk SUBleTLED IN PARTIAL FbULFILI:,,'ŽTOF TEE RRQUIREENTS FOR THE DEGR 2 OF BAOEELOR OF SCIENCE at the Mi'ASSACHIUSETTS ISTITUIE fi OF rTE'INOLOGY (1954) . ** e.-4e .... ..... 44 Siznature of ,uthor.... Certified by....... .. . s adSK . FTIO . ..... .. .. : Thesis Supervisor C . a.r *a * tt e * -* - ** o *rde a * * **o Chairman, DepartmentOsrnmittee on Underc-raduate Students : - 3 .,..@ . . 4 !. j; * , ACIOTiva iGD zil NTS The author wrishesto thank llan J. Faller and Professor Victor P. Starr or their ex- cellent supervision of the theoretical and experimental ork that co.xrise this thesis. OF C OUtIEIS TiB3 .icinowvledg ment s -zbstract 1 TIntroduction 2 The 1otion o a Circular Disk on the Free 9 Experimental 10 o3ries 1 17 Series 2 24 Conclusions 24 Suggestions for ±'uture Work 26 Bibliography urface VJork FIGURS Page Figure 3 1 Fluid velocities near the disk 4 2 Profile o 6 3 Figure 9 4 A sanrle disk 10 5 Coordinate sstem 16 5a Trajectories o 21 6 Free surface flow lines and lower level 22 7 Free surface vorticity pattern relative velocities ch-an-e o illustra..ting, variable relative to the dishpan disks a and b relative to the dishpan ronts at to nd lower level fronts at to 23 8 Free surface divergence pattern and lower level fronts at to TABL2S Page 12,13 Series 1 data for disks a and b 19 Data f1rom 4 Page 14,15 r and diskrs of 3eries 2 at to versus t. Series 1 data .i metIhodfor determining the vorticity at tie free surface of water in a dishp-an experimLent directly fro. the rotation of small circular disks is analyzedtheoretically. In two series of experiments it is folundthat the vorticity distribution over the free surface of water closely resembles the distribution at high levels o the tmospere. y using 'the silxmlevorticity equation, divergences ae calculated directly from the vorticity data. The signs of the calculated divergences are reasonable; themagni-tudes, -owever, are often large. t3yconsidering only the sign o .!ence, a close resemblance between the divergence the ree surface is demonstrated. in a dishpan the diver- atterns at and at highi lev-.ls in the atmosphere. -1i dea The model of the of develo -ing an r-3>-:-err-'.nta' atm;-oshereihasbeen a It might be of the useful be observed ortunate one indeed f.atures J o for meteorocloy. to ist some ~- omnt .orthwhile at this erly s oosed 1. T CiRODU CTI ON an e-r.perimental nm-odel. to the atmosthere, dynaiic rocesses an ontinuocusly. Dynamic processesthat would take week-sto obseilve in the atmosphere can be observed in a few minutes in an experimental model. 3. The three dimensional nature o the fluid motion is much clearer than any system of weather maps could possibly convey. 4L. Conditions in a model cn theoretical assu=ptions. often be For examle, the ade to it ssmapotionoften made in theoretical ,-orkthat the density is constant can be closely a 5. roximated in a model. Continuous measurements can be taken over sorme complete region of the model. many vi.riables atmosphere 'at The consequence of this is that cannot be measured with accuracy in the can be easilymeasured rom a model. In connection with the last feature above, most continuous measurements have been taken rom the dishpan experiments. :-easurements of the u and v cormonents of the fluid motion ave been madev.ith some accuracy. In theory, at least, r.1anyother variables can be calculated from!these velocity measurements. In particular te vorticitr at the ree surface might be w~tn1elal<;ed^ ol a1t '-{8'tL C ay au ?Fuchcomputations are both laborious and not very a.curate. -in 1%22t oW ihaft has the eer sa-.d abut v '-,aslrementsit mistt be tLouht tht values o shlould be ossible. o,"-,rfth. u and r re:,oably accurate The reason 'for le innaccuracy arises -frollthe definitiono fLte vorticity itself. T'0 calculate fro te definition te .derivatives.must be approiL.matedby finite differences: UA the Nowthough the percentage error in u and v is small, the absolute value o th1e error is of the same order of -:mgnitude asAU ndAV .Thus the error in is of the sde order of magnitude as itself. In a uantative analysis of f'luid motions, r accurate estn Ues of the vorticity ared'esirable. ince the calcula- tions of +he vorticity ffo,:h the velocity co:.,monents u,v are tunsatisf.acory found. some oa. er meeansof determininq This pae ex ll 1 must be ~zine a newr -nd more. direct of deterraining the vorticit-y in method dish pan eeriment. 'TEREM:_4OTIOOF A ClRCTUl:L? TDISK1M N 'jT F=' 3U':'OT J' Ifa luid in a plane is in solid otation than the vorticity is eual to2Af, ,.e eef is h'e angular velocity of -the fluid. it is possible, then, to hink of t vorticity in trms the o aofrotatiio-n. The followin derivation hysical connection bet::een vorticity era bly clearer. w^ill mnke and rotution co.nsid- -3Imasfne a cylindrical disk of radius R, tickness . uniform density oup ose h, next that the disk is immersed in a fluid contained in a rotating dishpan such that the axis of the disk is fluid. erpandicular to the free surface of the Erect an x-y axis system fixed relative to the dishpan and tangent to the free surface. (0,R). Imagine the disk to be at Figure 1 illustrates the orientation of tne axis system. The fluid velocity in the immediate vicinity of the disk will be given by U: Uo.,- (1) V=o The term ttimediate vicinity of the disk" will be clarified after the derivation has been coraDleted. 4~~~~~ Figure 1. luidveloceities near the disk r=a The fluid vorticity in this region is Supiposelastly that the W relative to the isk is rotating with ngular velocity ishpan. To determine the motion of the disk the law Of interaction between the fluid and the CAiskmust be kncwn. I shall suppose that the law of interaction between the fl uid and the disk is as follows: The force exerted on a unit area of the disk is proportional to the velocity of the fluid relative to the disk. he velocity by the resence o the disk. of the fluid will be assumed undisturbed If the (2) roDcrtionality constant is b then the shear is (2) '~b%/teLX Vye1 Vces (C= ut .v+ One way to interpret this shearing force is to think of the fluid as being undisturbed at some uniform depth do . reLis then the velocity gradient betwveenthe undisturbed do fluid and the disk. The shearing force to the veloc.ity radient. becomes proportional Tle assumption of a viscous shear proportional to the velocity gradient is often made in theoretical analyses of the viscosity of fluids. In this light the assumption of (2) seems reasonable. The motion of the disk can be separated into two kinds: translation and rotation. Each type of motion vrill now be ex- amined separately. I claim the velocity of translation of the disk is U.= Uo +o.R V=o To prove this from the dynamic assumption (2)VieL of translation must be knon. VY t. =0 Uve L (3) U 0 Cali vsiR UtyeL = Uo+boan - l ,,- RI R AsL is plotted below in unbroken line. Y. Figure 2. Profile of relative velocities -5- From this relation and -'iure it can be seen that the relative velocities of translation are siraetrical with respect to the center of the disk. In other words, for every force element acting from y eual to y eual equal and oppositely y equal Eacting force element in the region from to y equal R. Hence, with respect to translation the disk is in equilibrium. of translation ; Uoa for o 2R there is an the disk UO-O-R It is convenient that the velocity (in equilibrium, whilichI assume) is is from (1) exactly equal to the unrdisturbed.velocity of the fluid below the center of the disk. Jince the disk is oftoonstant density, any rotation will occur about the center of the disk s axis. result that te his leads to the useful center of the disk w-,ill remain atop the +water particle directly beneathit. There might be a uestion rotation as to v.teher an arbitrary could upset the translational equilibrium. 2 illustrates to igure purely rotational relative velocities (in broken line) at points s=netrical w.ith respect to the enter. .ince the vectors are equal and opposite (and it is always possible given one point to find a smmetrical no net translational force. translational point) there is Thus rotation does not afect o quilibrium. <lthough equilibrium with respect to translation has been established it is by no means true that euilibrium to rotationis concurrent. with respect This is especially true i an arbitrary counter-clockwise rotation WA is imposed on the disk. It will be seen that there is a particular value oC -which euilibrium occurs. at The relative velocities arising from pure translation are rom (3) U-teL=a('SR o Nyel_-\t.e It is convenient to resolve U*eL into components along the radius vector from the center of the disl,and pernandicular y iJotethat to this vector. Ž9ce (U.I l - - J _ · V - R+Tsih1S _ I 0 Figure 3 The result of this change of variable is that = oYsinecos U/TeLr (4) C/yeL= 4/ eL8 0 $f U/eL 0 due to rotation (5) O is =--C ButUjrwill not give a torque about the center of the disk so that the net torque per unit area is due to the relative velocity eL + /reL6, UU - (6) then (7) d ( 2 an Y'(asi ) aoi O (asiA T"= (9) which fron (4) and (5) becomes 4 A 6*-JA W+A) CIYJ S I-bY(asht9w)dr4 '4'+ rYcJe c wfich interates Irb R.+4) to -7The mo:.en ..thien - o of 'the diskc inertia dw -I (10) R is leads to th-^edifLerential euation ( - The solution of this differential e-uation is (11) W( )e W.- f + j w. =w t The interpretation of this e:uation is that i a dis is reeased an.glar in the luid at the the fluid beneath the disk. In equation (7) there is the tacit fluid acts only on one on both , then its approaches one-half velocity eonentially vorticity o W ngular velocity ac? of the disk. aces the torcue must b ssumrtion that the If the fluid acts doubled and eouation (10) becomes (0a) (ll)becomes =)and - U.~. i =-t (1la) WW.= Up to now no mention has been made of the torque acting on the side area of the disk. Vether the :neglect of this n the sid.e area of the tortue is reasonable or not depe-ds -c on disk or directly smaller torque on the side and h. can be mzadeby making the side a.rea smaller. The purpose o the preceding analysis was to ga-her physical facts for investigating the circular disks o -,easure ossibility of using small the vorticity in directly from t e rotation of the disks. be calculated (12) some ishpan eri -iments The vorticity would rom the relation 3= where W is the rotatin dish-oan. dishan. angular velocity ofthe disk relative to te U~~~z: -'C)- -ill tt aissu3 iOn f -aeziona,.bLe first I'.2e b e-(1ined he is tuhatof the assuijmedluidvelocity distribution. w-as originally ostulated as eistvi1g in the 4iimmediatevicinity of te disk.' But in the derivation the distribution distribution as mathematically used only in the area covered b'y the disk. Hence, the area o by the the disk is trhat innmediatevicinity of the disk." s meant This aea can be made arbitrarily small by merely making the disk arbitrarily small. Thus, other things bein eual better values of can be deterrLinedwith smaller disks., If the disk is small enough'the assumed velocity distribution will be closely ap-proximatedw-ithin the area of'the disk. The meaning and validity of the postulated lar of inter- action between fluid and disk was discussed when the law was proposed. although at a point in the traiectory o the vor- as ticity of the fluid may; qual twice the angular velocity of the disk, at some succeeding nearby T-ointin he tr.aiectory ,?here the vorticity not L.ave austed to te is differe the nt disk ;-ay new voraticity. According to (ll) place exponentially ib)diminish ., b, the visosity - ~rom (11) there are three .r.ays to hasten (a) dilinish h, tihe thickness of the disk, the adjust-ent: irinshi~n-h the adjustment takes the density of the disk, 'and (a)inc'rease of the fluid. perform -ofi:'nd -reduces L ~ ~ ~ the .. It can be observed tlat t.:ro funtctions: it .. -3terLsthe on the side --~:..e 1r~ue r-ca+ U. h z e '"~o adjutment -oeat k eikdisk. ',C.: ;;.:.{.. ... o IZ; w^u.J. 7J~~~~~~~~S -se Sof-r ::, To rin.:i _ , sll -'::izitude he ima o ...e i st term in euation (!i), the disk was ut firom.a 2lin sheet of ?,::,:er, the series of data the disk Disks ensity ",as 3i of radius of W...icl we;:re:etaken. 2-,v5s uch less %?an one. In the first series the radius of znd the radius of the dishpan 10man wer-e used with in the second series. Dwo was 150u. a dishpan radius of 300mm The velocity gradient does not vry as rao idly over te freesurf'ace in a lar)ge dish-an so t-hat 1ar ger disk radlii ay be used. The data in both series -was recorded by camera. camera mounted three feet above the ree sur-,e and rot at ing withLthe dishpan vWas connected to a tiaing mechanism every second a .otoraph A so that was t.,:en of the d shlan with be disks floating on the ree surfiace of thew-ater. By projecting the pictures on a grid in olar coordinates data was obtainable directly rom the neg:ative Thne disks were numbered LramCs. or identification as illustrated in figure 4. Figure 4. A sample disk The position of the disk is considered in relative to the dishpan. Thllecoordinate sstem olar coordinates is shov in -10 eunit, , r i,- been L-trs ad C unit o i G.-, n. £aor ct0r 1Lstbelie d to .rinsforr: r to cr. Thle convers ion factor s M±ffrent for bot, series o lata. , :ie ,,,easured in dec-.-ees. =',1:ura5~e e wad...Q.dJL Kd af tter 2re valTues oJIte var i&2e z sc on ds hv e ea r se . ~a~ c . avee been by using, central differences. _s X ;esti.'mated from.t-e discrete data For e:-:mcule, O+A 4°-. Ahrln Figure 5 Coordinae stst.. sstm e rrati i ' o te dishpan Series 1 In seriesone the rdius o te thie radius of tiie disks was 3. to te the dishpan v'as 150m Te fluid otions relative dishpan were induced by heating the lateral an w-ith amipns. lThean,:-ular velocity was 20.8 degrees sec'-(3.47 r.p.r, and - of te 0.363 rdians ameaof dishpan, sec-l.) , -1 - Table I contains data -from series one or t=,o ais';.s, a and b. Here 4.81 units equal 1 cm. Graph I is a plot o disk a. Grah versus t r versus t and II is a similar plot for disk b. or In addition Li-ure 5a depicts the position relative to the dishpran two disks lotted at three by points econd intervals. Graph I indicates that over the one minute a 'l1aspredominantly negative vorticity. that there are large sc.le changes o scale changes superimposed. of the eriod disk It can be observed vorticity with small Such a attern disk is moving in a meridional path. is cos-ion when the (A meridional trajectory is indicated by an approximately constant value of r.) Graph II is much more interesting. I, there is a -eriodio variation is almost eactly of r being values o r. As op-posed to Graph The variation o in the same sense as r: increasing values conjoined with increasing values of ; decreasing of r conjoinedwith ecreasingvalues Off relationshipbetween r and had a significant eriodic variation o Since the gradient o apro.-imately was observed for every This dis-kwhich r. the vertical velocities is zero at the free surfacethe vorticityeuation below can be assu.ed. But is constant in so far as e-perimentally possible so that (13) Qualitatively, this are sociated eans that increasing values of 1with convergence and decreasing values of -12- I I I I b I I. l I Table I. Series 1 data or disk a. 4.81 units equal 1 cm. R .A_ 20.8 degreessec.1 I I 150 mm. I I I I I i I Al Table I. Series 1 data for disk b. 4.81 units equal 1 cm. R_J-= 20.8 degrees sec.-1 is I1 150 minm. - i - T(uclits seC) *o -/t Oro 60 to *'0 --0 so c0 - /0 GraphI. r and ve(rsust. Series 1 data for disk a. -r d ; T(-nits' -5(J st'SC-/I f /00 to 0 w4 II. r and-versus t. Grapbh Series datafor dis b. - Figure 5a. Trajectories L6 - o diskl- a and b relative the dishan. Deries 1 dta. sec. intervals. to Points at 3 -17with divergence. Since graph II is typical or a eriodic variation in r, it can be concluded that there is convergence to te of a troughaand divergence to the east of a trough a free surface o the water. the This conclusion is eactly to the divergence pattern observed at hit 500mb) wiest levels (say analogous above in the atmosDhere. Following this result a more extensive picture of the vorticity and divergence fields wEs sougat in Series eries 2. 2 In series 2 a dishpan of radius 300mm was used with disks of radii lOnmm. Heat was supplied to the higher radii by electric light bulbs mounted directly below the outer region of the dishpan. Sixty disks were placed on the free surface of the wlater with an initial attempt at uniform spacing. using a technique developed by Alan F. Faller, potassium permanganate crystals were scattered over of the dishpan with the heaviest radii. the floor oncentration at the inner These crystals dissolved to form a dye which remained at the lower levels of the water. After several minutes elapsed the dye was observed to be concentrated at the inner radii with a very sharp boundary between darker and lighter dye. Temperature measurements have shown that the darker dyed water is about 5 cooler than the ligvhterdyed water. Analogously to the atmosphere distinct outbreaks of cold water move to higher radii. The ensuing fronts appear as dark bands progressing southward (toward higher radii) and/or -1 - higer vales of e eastwara(toward . for series 2 was held at 7.56degrees sec -.IL (1.26 r.p.m., 0.132 radians see-1 .) A series of data ten formed is contained in meaning as in 5.36 l units equal 1 cm. hen a strong cold front had r and 9 able II. ¢¢ , however, Iable I. - have the same is not directly entered into Table II. The vorticity and divergence pattern to . as required at t_3, t, tl Data were compiled fort and t+S , where the subscripts refer to the relative time in seconds. Estimates of CW_ and W.+1 were calculated by central differences and are designated -40 and A+4 · A¢~4J-3 A mean value of Kh and A1J (15) taken as W 0 . s°+.. i- E7= The vorticity ~ E-7 is , is given approxinately by (16) 3 ZA§° The absolute vorticity is f plus2.(Vwere f ecualsZAL An estimate o fIt ororalternativebL~ atraiei,9 - is calculated by central differences and is labele.: (17) ~afte ~~( ~~~) _ '_¢-_ _ Lastly, the divergence from (13) is approximated as follows: 1 out the constant factor two. canceling -19- A3 -,. Z'" . 1±o2 / Asz .o/. f.= 1.C //yZG*04.3./ ~~~.. a.im /d -.,. I-.-s=' , _IS _~uv~itS V - * .|- _Zd4L *1;-~ />1*: t- _ifd-. ZZ•11~~*. ~~3~~Z ,,iwl. , _ I,'t ,-- _ 3±A 4 Jmk - d. -- 0-I- +-. -°.60 - -.-/~ . 2z / dI -*' /. ,4.0-., - +.1 - L~ .../i .ii A&, , r-..t / 3/$. .-- ._aP Z - ~-~o, : _d,_ -$/- ZQ0, - . -a . e.± - ,f± / Z "_ . ./3L+/' , _4Jo.I -... z'-. --- 6/- _ -- 2e. .s" - ./ r. . .0 -. I -re. 7 . f - of -,-,Q /.4 '- + -31 .I -2.o -4-." '--~, x.3z '.-S..' J. .1z-.'-.~ 9P -33/ azL-.I JI.. /.,¢' -.," -- -d., -7./-~_4"i~ --. ,dme ,,a.L -Z,,, A.--,,/ ed, --3, 2. + JJ.a6 +.:Z37 -.. 2 . -- _+'Z 3*-d' ZV.-.- d /~. /L k -,/a.Z.e ''.~F ..4 2:232 _/-J, , .a. 3. -3. / _ ~ j2_ . z. Z'/2. ~~J.A .I / ~~z2 307 -~~ /./ . -. ZJLR -.. , _2 .,,;a, z ,J7 --D. '-/. ~ f/. ,~ 31, '; J9~~.~ *0 7J- -,d. / ~ --. Zap -. . -- £a/,fz -:../.--./f _~~~ ~' -'--,t -- '/2Z_.E -~. 9 /Le .~~/ .A. A y-/ ~ -IL. a - 7 +'3.#1 ~-&d *4 .. 4 72.. 1 *3/ -. .4.0~ ,..,,d%.7 *4L - -i3 9 0./.Z_/. & ¥.t_7 z 1,; P~ /-- ~_; - ' 'f',f.gq4. +la+3.3 J 0-2 ~f 7-- -/z -/.0 -! z _ '/ z3a- . N Z . --7 --4- am. fg ---o& .f, Table II. - f , 4,7.d ,. 2/f - .7A a- o. / +/,/3 a.2. +/7-/ 4A~, o -13,P a.77 F +,/, , /1, A, /.l *:; Data from 40 disks of 'eries 2 at t. -L 7.56 degrees sec. 5.36 units= 1cm. R = 300 ', mm. -20Table II ~ original sixty contains the information ,diav, for forty disks. ' X K -I 'W ,+ Tenty of tlhe diskCs were either in collision or rubbing against the outside rim of the dishpan at to and hence were not used in the calculations. igure 6 is a map of the free surface flow lines and lower level fronts at to o. igures 7 and 8 are maps at to of the vorticity and divergence patterns respectively. The prominent features of figure 6 are first the four approximately symmetrically placed high pressure regions; second, the two fronts in the second and third quadrants; third, the three sharp troug.s; and last the strong polar vortex. At this point, one serious disadvantage of the experimental procedure is apparent. interesting This is the absence of disks at ositions in the dishpan. Tor example, it would have been desirable to have disks in the upper level trough in quadrant three. In figure 7, VLonewere there, unfortunately. the vorticity pattern resembles strikingly that of the atmosphere: Over the four high pressure regions there are maximum negative vorticities. :&aximum positive vorticities occur in the major troughs. In figure 7, atched areas represent negative vorticities. The divergence pattern interesting. s depicted in figure 8 is However, the magnitudes of some of the divergences are much too large. This must be due to the fact that the )0 Figure 6. Free surface flow lines and lower level fronts at to. Figure 7. Free surface vorticity pattern and lower level fironts-at to. lhe two broken lines are characteristic flow lines at the free surface. '? C} Figure 8. Free surface divergence pattern and lower he two broken lines level fronts at to. are characteristic flow lines at the free surface. -24- err-or in V is of the sme order of mriagnitude(or greater) as 9. This can readily be understood when values of f are examined in Table II. Values of S9of one, two, three degrees cor.espod approximtely to the error in V so that the magnitudes of the divergences must be considered to large error. Nevertheless, the sign o probably reliable. in igure subject the divergence is To emphasize the sign of the divergence , regions of convergence have been hatched in. Again, as far as the signs are concerned the resemblance to the atmosphere is strong. The eastern sides of troughs are regions of divergence while the western sides of troughs are areas of convergence. In the atmosphere this condition exists at upper levels (above 500mb). C ONCLUSIO NS The vorticity and the sign of the divergence patterns on the free surface in a rotating dishpan as calculated from rotating circular disks closely resemble patterns at a high level in the atmosphere. The divergences are much too large, although, due apparently to small errors in measuring the rotation angle of the disks. SUGGSTIOSl FOR FUTLE WORK It is ossible that by a proper choice of disks, vorticity and divergence patterns at other levels in the dishpan could be measured. Experiments to det;:rminesuch patterns by using liquid layers of been successful. slightly different density have not If the magnitudes of the divergence are to be calculated wiith accuracy a much better method of determining the disk rotation angle must be used. Insufficient time prohibited experimentation to determine the best feasible method of taking data. 6ome means of locating disks at interesting positions would be desirable. It often happens that an interesting situation occurs with no isks near enough to give data. An increased number of disks would give an increased amount of useful data if the number of collisions did not increase directly with the number o collision, its data are useless. disks. I a disk is in One third of the original data had to be discarded because of collisions. -26- BIBLIOGRAPEY 1. A4lan 3. Faller, "A cuantitative Study of Eperimental Models of the Atros here, " S.M. Thesis, . I. T. January 19, 1953. 2. Dave Fultz, On the Possibility of Experimental Nodels of the Polar-Front Wave, t 9, 1952, p. 379-r, Society. 34, Journal of ieteorology, Volume The American Meteorological