R.A.E. Ref: Aero. 318.R/166
-In S&aplano Tank testing it has been customary in most tanks to make a correction for'the air drag dn the model andtowing. apparatus when making watcir resistance. measarernent. It has been found that there are sometimes large effects duo to .ai.r flo* thich
are not eliminated 'by this procedure. . The .amiil1e evidence is
sumniaried and discussed. in this note 'and reconrriendatioth zire mad.e
about the pzoceduro to be adopted. in tank.tests. Contents
Appendix on measurement of air pressures in tank testing.
[ak. v. Scheepsbounde
Teclrnische HogesccoI Delfi by J.P. Gott Ph.D -1-/ 1. Introduction 2. Method of testS . Presentation of results L.. Sunderland .. ..5. Conmarison of modal with full scale for Sund.erland
6.. Comparison of model ith full scale fOr Lorwic-.
7. Comparison of Lerwick and. Sunder].arid.
8. Empire Boat
9. Barracuda and. Spitfire on floats
-1O. Compa±'isbn with German tests
Ii. . General Discussion .
-- I
Class No. 533.651.6
Technical Note No. ero.iL6O Technical Note No. Aerà.116O
Ju1r, 19L14.
ROYIL AIRCRAFT STLBLISv1ENT ,,RARNBOROUGE
Interference between Air.Flow and Water Flow in Seaplane Tank
'¼
S
Technical Not& No. Aero 26O
I /
I . Introduction
In the noril routine pf water resistance measurements,
as it has.
.isuallybeen done in Seaplane Tank testing, corrdtions are aDpliod forhe air.forces on the model and measuring apparatus'. To detexinc
these corrections the mdel is towed just above the water surf.ce .nd
the air forces and moments measured ii
this
condition are subtractd,from the dorrespondin qunntities measured with themodel in the ator., The residue is-regardod as due to wate-r forces only. The borrection
to drag is very large -at the h-igher speeds but. the pitching mporaent
corroctjon is usually, small enough to be. neglected.
'in a repo±'t on tankcsts.o± a model of the pir Boat.1'it wa
shov that the farces on a,iOodel Can be largely affected by the
air
flow round the model and that the
usual correction for air forces is
' not suf'icieit togive, a- unique result for the water. forges on the model. This When spoi'ier ridgswere fixed to the model, to distuit
the ir flow round the afterbody, large' changes were, produced in the pitching moments and there were also changes in the 'min±rnum 'watei
c1rag nd the angles for minimo drag. The spoiler rid.gcs d±dnot
affect the easured. àori-cbtion for-air forces. These eff'ets.were
attributed to a--low
air
p±'ese in the gap between the afterbody planing bottom and the water. It is not proposed to consider thedetails of the,air flow in this note, but to give d.ta on its effect
ona nuiIer of modclwith particular reference to water drag. The low
air pressures ha been mea9ured directly in one
case, and the rosi.ts'
are given.in'thb c.npenclix.
A low aiprcssureunderti'e
ftcrhocI-affcts pitching moments directlynd has the additional effect of
surrounding the 'afterbodywith spray and broken water. It isprobabJ.y the friction of this water-which produces most of the effcctswhich have been obseed on drag'.
The discoveny of the air flow effect on water forces s given
-part of the solution to
-a dliff'-icu.lty in.wator.re'istarice measurementswhich has become serious in the last few years. For exaole the
Sunderland although originally 1esigned for an- overload weight 'of about
144j000 Jb., has been flo off at weights in excess of 60,000 lb.,
whire
the original tanks tests would. haie given the limiting weight as ebout
50,000 lb. ,Detailed, exorLination of the original resistance measurements
shbws "sticking", not at the himp resistance but at a second water
re'sistancc maximum or hump oouring at pecds near take-off whore,
.
- although the water resistance -is less than at the hump, - thetotal
resistazce with air drag added exceeds th
thrust.
This second uirp has become, usual. in tests of mien designs but it has always, beensuspect first, because there is little if any
full scale evidence for. a sepoñd. hthip an second, because".it was found that the new tank
carriage 'ad boianc recorded a bigger huap than the old carriage and.
balance. A rough preliiinary test was rde to see whether this second
hump was associated with air flov. The Sunderland model was xjn under conditions to give the high resistance at high speeds end a drag of
3,100 lb. was measured. Spoiler ridges were then attached-4o the
/
model 'and the measured- drag was reduced to 2,070 lb., a difference of1,330 lb. or 6j% of the mallOr value., On
the basis of this reult it
waS decided to investigate as 'many nodels as -possible'to find the -
-extent and ±aportancc o,f the effect of air flow on water drag. Thern work was done at various 'times over a period of about two ylears. Most
of the tst wexe done by
?Lr. C B Baker B sc., ki,r. H. B.11ett, B.oc.(Eng), aiia i.r. D.I.T.P. Liewelyn - Daves, B.A. -'
Tc-cbnical Note No. L.ero.i)GO iethod. of Test
The method. dopteci was to run the model behind a 'screen so that the airf1ow past the model ws' reduced. to a ncg1i;±ble
sped.
Thearrangement of the
scren
is illustrated in fig. . Testcan sill
iDa made in' the original way vth the scrc-en raised. In the region
near the model the air velocity, relative to the carriage, is reduced
to about ift./sec., at a carriage speed of
32ft./sec. and is irrelar
'in direction. This is sufficient to reduce most air effects2 depo g on 'the square of the speed, to negligible proportions. No air flow
cortions are applied to the results
of tests behind a sc'ecn.The use of a screen in front of a tank model is not an ideal arrangement and. a number of conflicting factors 'had to he considered in deciding to adopt the screened iiothod of testing. Thu whole question i's discussed later in this note but one aspect should be
noted hero. Tank carriages have been designed vdthout any reference
to .the air flow in the position occupied by a model on the balance
and., as a result, in the R.A.ES tank the air vclQcity relative to the
nodal is groitc-r than the speed of the carriae:e. This exaggerates
air suction effccts on resistance models, coaiared 'oith dynamic models
which arc tcstd in el::ost undisturbed. air and. makes soe arrangeuent
such as a screen essential. Presentation of
Rsults
In taril tests it is usual to'.plot curves of pitching moment and.
Lter drag against attitude for eah of series 3f speeds. From these
curves the drag. corresponding to. ny gicn triiming moment, to mininom
drag, or to any assthoed take off attitude, can be obtained and take-off drags platted. against the speed. In the resent tests these curves are
available, for a nuLTher of models for tests both 'ei.th end ithkut screen,
so that the complete collection curves is vexy extensive and. would
not he casily.intelligible to the
:eneral reader.' Hence this note'.gives only selected cures dosi.'noci to illustrate the important points.
In most cases the take-off .irar' corxesDonin:r to,,an assumed. uniing
attitude is plotted against tie snced.
Mod.els of' the Sunderlanci, -.Lex-iick and. mpire Boat and of the tvn floats for Barracudi and
Spitfiro are considered..
-Sunderland
-Tank tests on the SundCrland model are nore extensive thin on any
other tioclel and. can be uned to illustrate' most of the moints which
arise. Pig.2. gives take-off drags based. n an assumed running
attitude of 6° hull datum (1 20 wing chord) roar take-off and.
free-to-trim at lower speeds. Curves "screened' and "unsoreened' are given
end they illustrate, the reduction of drag near take-off speed which
can ho obtained by screeuing the model.
Fig. 2k clo illustrat5s the
difference which i soietimes obtained between the drags neasurca on
tw carriages.
The, screened. tests gives the smallest drag and. the'new carriage creonc the largest drag. The old carriage (unscrecned)'
gives values close to the screened tests, except near one speed. where'
they rise to tie highei values. The differences are attributed. to the
diffe'ent air flows under the tvo cariaes. On the old carrihgc the
drag balance is situated directly ahead of the riodel which consequently
runs in 1isturbed. air and this gives results which, while,
they are ajito
definite and. repioducd.ble as experimental results, arc erratic in the
sense that in one case 'they arec with the l4gh values of the new
carriao
unscroened. nd in another, case they agree !th the low valuesc the screened carriage, while they Can also have internodiAte values.
-3
Technical' Note No. Lerol)+60.
Pig. IA illustrates the relative positions
of'mod.el and balance on theOld and ne carriages.
An indication of this result was obtained some years ago, without
being properly .und.ci'stood, as the fbllowing auotation from reference
2
showth:-"In the R..r..E. tahk it has been found that changein the apparatus under the csrriage soriet:Les produce. approcibIe changes in the mc-asuród. drags which are not entiraly eliminated VTlCfl the appropriate air drag corrections arc applied. It seems probable that it will only be possilc
to obtain identical results in different tar:s using
dif crent types of measuring apparatus if the -air drar corrections are eliminated bycreening the model".
Corresponng to the ca-ngos in drag illustrated in
g.2., ther
arc changes in the "cross curves" .of drag and pitching moment againstattitude.
Consideration of those curves is usually not vo instructiveend. fig. 3 is sufficient tO.. illustrato a peculiar of±ect hich has boon ôhdned. on 'a number of occasions. Fig. 3A givQs curves for the
Sunerland at L4,600 lb. and. corresponds with fig.2A. Thee curves arc for the screened model, togethOr with three curves for the unscreenod.
eodiol. for comoarjson. The curves for 60.0 and 690.'L:notsBhow the
red.uctj n in drag which hats been obtained by screoning the model but thei'e are attitudes nO-ar which the dragis not. much changed and this
gives localiscal peaksin the curves. It has usually beeh coni4ered suffiOient to plot thcse cives from mcasurOmonts made at 2° intervals but this is inadequate near the peaks
end recent work s.gost_they
may be even nc-re localised, than is shown in fig.3A. Fig.3B gives
similar results for the Siftickrland at 56,000 lb.
The drag curves of fig.2. show a small seconcJ. huip just before take-off on the screened. model. This is because the assumed. take-off
attitude, 6°, has coincided. with one of the pQal:s in
fig.3.
. Thistake-off attitude was usc-B. to cor:parc with fomer tests on the Sund.erland
model. By takin-off at a slightly grcater attitude end lower speed the second. can 'cc avoided.
These peaks, in ti'iC drag curves seem to correspond. to
a critical
condition in which to relativa notiin,oI the ator past the aftorboa.y
planing bottom induces an air flow and. a sucton 'even on thc scrccnecl model The followin- ob ervati ens arc in agreement with this
siiposition. . .
At attitudes above and holo the 1)Ca1c the water flow was -smooth and the afterbody was
ffctively clear of the water.
At theattitude of the peak the ...etor surface was broken and. the afterhody, altliough still clear of solid. water, vas surrounded by spray.
Tests vrere macic on a forebody without aftorbody and results
obtained which agreed fairly closely with the curves obtained by
siodthing out the peaks as indicated by the lotted lines in
fig.3.
Attempts were made to sDoil ho airflow in the ga between the
aft:erbod.y and. the water by means of spoiler ridges fixd on the olani.ng bottoL'1. This was moderately successful. .To be effective the ridges iadito. close almost the whole ga between planing
h tt.0a arid water, so that the ridges were scraping against the
-5-Technicc':1 Note No. Acre, ii6O,
These peaks and. the associatd. second hump are a prominent feature in
tests on aSunderland model but, as suggeted aboo, they are a phcnomebop depending criillyon the conditions and. very
in tests of other apparently very similar models, thuc i
littl9 trace
of the effect.
For the' und.rind model the screen has littld effect
on pitching moment. This is in contrast to the E::pirc oat 'model where
the effect of air flow on -pitching moment is large. >
-5. Comparison of model with full scile for Sunderland.
Tako-of times at weights up to 56,000 lb. arc available for the Sunderland. and it is useful to compare these with the times estimated from the-model in the roi.ttine way. As already stated, the original
trtk iests gave a limiting weight of about 50,000 lb. and ggested.
sticking at high speeds. It is now ifloVn that the original tests -and th estimates based on ther. requi.re correction for tc fo.lovdng three effects. '
-(1). Screened models should bC used for the. drag measurencnts.
The tank tsts depend on an estimate of wing lift with slipstrais
and ground effect. This cias und'erestinated compared ith' the deta.
nowavailable.
The air drags used in the original estimates were ovoitimdted by a factor f abOut 2 compared with values since agreed. upon.
-Any one of these corrections one is -sufficient to reduce the totul
ostimated drag to less than the
thrust for aieight of 56,000 lb.
and the three cbrrcctioris together result in reasohableesimatcs for
thQ take-off times.It V-las necessary to rake 'new tank tests using a screened model nd
thebest values for the wing lift..
Some of the results of these tests are 'givon in fig. 44. The assWeod attitudes to which these cuxvescorrespond are free-to-trim at loe speeds and the hump and. 60 at high
speeds. Three oiñts may be notOd in tiesO cu±vcs;
domparing with fig.2. the -mot noticeable difference is the lower takeOff sp'ecl d.ueto increaeci wing lift.
Three.cues in fig.Li.. show the second ht effect, although
reduced compared with the unsci'eened model.
The eurvOs for 50,000 and. 56,000 lb. show an
irre2Jiarity at the
himip compared with the currcs for 58,000 and. 44,600 lb. T}is
was definitely due to
failur
of the. step to plane at thesc loads until speeds of abo-it 2 knots were reached.t tha two lighter
loads planing cornmc-nced at ebout 15 knots.
-Take-off times according to various assuniptions were calculated from
the new tests with the following
(Li.)
Take-off tunes in seconds
-6-Technical Note No, Aero.14.60.
Column L gives t .e times calculated frm fi )4J. and cplumn B
sho'z the effect of ..smothing ut the second huir. The drags used. for coumn C 'ere obtained. by allowiixg for the triiing moments due to' thrust and. slipstrethn (elevator neutral) \estlmated. from reference 4..
This reduces the attitude by about, 2° at the hump. Colunn"D corresponds to C with the second hump smoothed out and in column E there is, one. takè.off tine calculated frdm the iiodel results .using renning 'attitudes
oberved full scale. Column F gives full scale take-off' tifre's and the. values in L and. F are plotted, in fib,. 4B.
-- Lt the heavier loads the mOdel times are too long. No one'
definite reason can be advanced to account for this but the following
corisiderations are relevant.
-Comparison of the figures in columns A -to D suggest that the Th.ctors there, considered cannot account for the differencç. In particular, at remains of the second ht. on the-screened model has only a .n1l effect on the take-off time.
Take-off times depend. on the difference between thrust and t9tal drag and at high loadings this difference is small compared with
either quantity, s.p that ii 1 errors in thrust or drag have an exaggerated. importance in the final result. Thus a error in either quantity would. accOtit for the whole difference at 56,000 lb. weight. It follOws that high accuracy in the final result cannot-be obtained when the take.off' time is long.
(3)' iil the model calculations assume a take-off ttitud.e of 6°
hull d.atuu (12 wing incidence) and this may not correspond with the pilot' s .ctions. He may, 'for example, be more inclined to piiJ.J. ofl as the take-off speed. increases.
No allorance" has been made in these calculations for any acceleration effect or for scale effect on skin friction. The foimer would.
make' take-off times even, longer and if. i is as large 'as, has ' sone,tines been suggested5 there irnist be some othereffect 'to
compensate. Samä work on boundai layer flow on tank models is in hand.
When theo calculations were macic, the ta1reoff times given in ig. 4.B were
himost the only full sca1 data available. /
-I
ei.ght lb. -A, Fic-to-trim / B As AThut' second huip removed. "C' Trined running attitude D As CThut second hump removedE'
F'.1 scale running attitude F' scale times. 38,000 44,6op 50.000 56,000 16.2 29.2 4.7.4. 86.0 28.8 1-6.7 b2.5 16.2 '30.4. 44.4. 83.4. -. 29.5 -3.9 80.0' 31.0 -19.0 42.0 67.0/
A knowledge of take-off distances and. speeds would have made the con-parison much more complete. Better information is now available and
it iriaybe' worth while to make another and more complete compa±'ison. Comparison of model and. full scale for Lerwik
Full scale take-off times are available for the Lersick ove± a
range of weights and-tank tests were made at these weights usin
a screened model. The resuJ.ts are given in fig.i which also includes
one curve obtained oii an unscreened model and. showin a second hump.
All 'these drags are based on an assumed take-ofi attitude of 90 hull
datum. Take-off times were estimated by the routine method. and. the
values are compared with full scale in fig. 5B. The agreement is good.
This estimate is, of course', liable to be affected by all the factors enumerated in connction with the Sunderland. tests.
Cdmparison of Lervick and. Suhd.erland
At one time the Lerwi.ck was reported. to have a long take-off
run and, to find whether this was due to
thc'hull, special tank tosts
were made in which the Lerwick hull 'was compared with the Sunderland
hull. To do this the 1/15 scaJ.e Sunderland mdel was tested as a
1/12.8aale Lérwick -model and. the resulting drags were compared with those measured on the Lrvick model. The original, results are given
in fig.6A, from which one wduld conclude that the Snderland has a higher drag than the. Lerwick at the higher speeds. Both models show a second
hump. These 'tests have ,ince been repeated. using screened models and
the results are.4ven in fig. 6B.
Hero the socóid humps have gone and.there is little to choose betcen the drags of the models at high speeds. There is no definite evidence on which to decide which of these coin-paris3ns is most nearly correct, but all the indications favour the
screened model.
The curves of fig6A were given in ±'eport No. BA.1599, May, 194-0.
'Several draft copies of that report were issuOd. but it was then found that the results were open to question and the final report has never been issued.
It is now of little interest.
Erapire Boat .
In the report on the Erire
Boat1, tests were made on the O.ginalmodel and on a. model in which the air flaw was disturbed by spoiler
.ridges. Attcnt-ion'was directed mainly tothe effects of air flow On
pitching znoraept. Only minimum drags' were plotted arid, the reduction in drag obtaind by the use of spoiler dges was not very great.'
A rat1er greater reduction would hav been foi..tnd if a running attitudc more suitable for flying off had been assumed 'at the higher speeds.
Tests on the Enipiré Boat, riodcl have now been reppated- using the screen
a±id sOnic of th results are given in fig.7 which shows that, for the
assumed attitude of 70 hull datum near take-off, the screen gives a
considerable reduction in drag.
In fig.
7 the screened model and the model with spoile- ridges give. appreximately tiae sanedrag, hut 'this is
not a general result. At other attitu.dcs the screened model usually gives the lowest
drag.-9. Barracud.aand. Spitfi±'e on floats
Results of test, screened and unecrcencd, for twin float soaplanOs
are given in fig.&, for the B.rraoud. and fig. 9 for -the Spitfire. Technical NOte No. Aero.14-60.
10. C.nnparison with C-erinan tests
vapor by Schmidt, on scale effect in toni-: testing, ivos
p.
Technical Note No. exo.14.6O
These show effects similar to those already disissed and. call for little special comment. For the Barracuda the excess thrust at high
speeds is small enough to inalco the question of what is the correct drag a serios one.. For the Spitfire the thrust (not shovn in the
figuro) i
so great that the exactalue of the rvr drag is of little
practical inrpQrtancc.results of tank tests on identical models in to CToman tanks; the
: Thur T ni (T-fsv) cnd the
crlin Tony (vs.
Thc krbur t'- iiused. screened models while the Berlin tank corrected. for air cira in
the usual. way. This gives a conparion between screened bncl unscroncJL
models and there are large differences in dra; at high seed, vhicharo noted but not explainod by the Gman author. hQdels of 1/5 nnd.
'-1/2.5 stale were used and. the differences in dra
at hil speeds can
be sulTunarised. as fo11ow. ( In co rin models of different scales
all drags are first converted to full scale values).
The two screened neaols .ive clesely the eame drag.
The 1/2.5 scale moctel unscreesed has premtCr drno than the screene,d. model.
(3) The. i/5 scale model unscrecnedw's less drag than the screened model.
Of these results 2 agrees viith our tests but .3 is different and is
an dffcct not fCunci in our work. Perhaps the most significant point is that screeied models of two different scales give closely similar
results while toreare lai-gc d.iffcrcnccs on unscreoned. models.
11.
enâraljJisusion
The test reu1ts rccorded.int
are best re;ccrdei from two distinctthe qustion of iniediat practial
screened imidl give-s a. reliable, est resisance and socond.there is the
of hulls in
cnerai.his -note ibad. to conclusions . which
eints of view.
first, there is
1nrorta.nCc, of how ncaly a
mate ci the full scale- water
viidor question of th hydx'odynamics It is convenient to consider the second question first. The nest
important :.cneral conclusion is that it is not Dossible to donsider the water flo as iidependont of the air flow. The real condition s a comlicatcd interaction at the surfa.oC. of
seuation of two
fluids, air and water, which the air flow has, in general, quite important effects. Effects on wato esistanco have been discussed so
far in this note.
I.t is now dusirel to draw attention again to theinfluenq of )air uetion on prrpoising rhich was the main subject of the former report'. Bothreodoi and full scale tests lead to the
-conclusion that efficient vcntilLtion of the aftorbody is necossaxy to
attain
high degree of stability.. This has usuLly been obtained - by the nse of a clee step. Recently air ducts in the ilaning bottomaft of the step have bben tried.
It is clear that ventilation
V produces stability by rcducini air suction effects and it seemá almost V certain that air .sUCtiofl effects providc; the explaat±on of the finite
disurbance often required t: startperoising on models.
It has
already been explained how air suction ffects can -roduce vcn localisud. pcak V the cross curves against attitude as. shom in ftg.5..
\
Technical 'TC No. Lcro. 14.60 \
refers to the screened model. Thc duos for the unscreencd model refer to the exaggerate&air flow conditions under the tank_carriage.
The conditions on a dyniric model will be sbrncwhere htwecn the two
-sets of 'curves and, although the shari5xicss of the pcais may be reduced, it is. reasonable to assume that there will still be 'a localiscdincrcasc in suction in the immediate neighbourhood of a particular
attitude.
-Considtor a model oscillating with a smaU amplitude, no that the motion is cicinpc1, and. then let the emplitude be iperoaned until it includes
an attitude atwhich suction effects occur.
If the suction effect
is
uf1'icient1y localised. it will 'act like an impulse applied at a particular phase in the oscillation and/ it is not difficult to show,from the usual eressions for a
npecl haruonic oscillation, that ±fthe phase 'of the 'mpulse is suituble the model ll then execut a
continuous undamped oscilation.
In general the period of the continuous non-sinusoidal oscillation
will be greater than the period of the duied oscillation.
This isccju.ivalont to the theor of a clock' s esceoment and the pase of the
imiiulsC required to make the period independent of the magiütude of
the impulse is a condition which must ho;stisfied in a clock.
ctually the porpoisipg problem is mere coni:olicated than this
since te
atitud.o will, for a large oscillation, both incrbasc and. d.ccro'asethrough the angle at which maximum suction o'ccurs ECnd poxoising will
d.eponci. on a difference between the suctions in tese two conditions. Such a difference will exist if, for example, there is a difference in phase between the oscillations in pitch and heave. Theiexistencc of
suction peaks producing jozpqising will not necosarily show in moment
mcasurOraents made tmcler steady conditions (drag balance measurements).
If t -here is a phase d.ifterence between the oscilJAtions of itch and 1eave the'critical relation between pitch and heaye nay not occur under steady conditions dt the same load, and. spCed. Data on which to develop
a thcoy on these lines might be obtained. from a detailed analysis of
porpoising oscillations, and. appo.ratus for this is being made.
Accdrding to this theoiy the essential feature is' not the disturbance required. to start porpoising considered as a force or a moment,- but the
ol!tude of oscillation required to reach an attitUde
at. which suctin ei'fects occur. An indication of tfle correctnessof this view'as
obtained on an unstable model which was made 'to oscillate at small steady
amplitudes by running th±'ough a long0and voir shallow wave. 'The never
the double amolitudo ±'cachcd about 5 , porpoising of much larger
nplitud.e commenced. The critical condition need. oply be reach3d. once -and could be reachc-ri full scalQ due to any number of chance circumstances
which do not exist at' afl under the controlled, conditions of tank testing.
If this theory is substantially cdrrcct the
model and full scale problems-are clearly indicated. The model stability limits' shouJal first be
determined in -ho 'usual way using an 'arbitrary large 'ciistuftancc and. ,then
the critical implituth of oscillation rcqui:
ed to start porpoising atthese limits should be measured.
In full scale tests the probability
of 'reaching this critical amplitude should be determined. on a
statistical
basis for variJu take-off and. landing conditions.
\
It ws indicated. in section
, that
'the exaggerated" air flowcon-ditions under the tank carriage make the use of some arrangement- such- as
a soreeti ssentid in testing resistance models. ' An anso:er must then be foutid. ,to h'c question, mentioned above, of hov nalrly this procedure gives an accurate estimate of the full scale wIter resistance because the differences bctrccr the results of, screened and unscreeneci tests - 'are serious. In a general way there is much to be 'sold in favour of
-f
using screened resistance models.. Consideration of. the Grnian tests 1as shom that screened ñiodels.give identical results on two scales of model,
while
nsenédmodels do not, and coiarisons with full scale for
Sunderland. and. Lerwick show that screened model results are reasonable in.
thos;cases.
An alternative procedure would be to test1resistance models under conditions. vhere the air flow is undisturbed as is done fr dmaznic models. (see Fig. IA). It would then Drdbabli be found necesary touse complete models, instead of models cut off at an+.arbitraxy deck
line, because the air flow is very easily disturbed by the test conditions.
Thoro is the further difficulty that tank tests
re controled byFroude' condItions and oit is imossibie t.o use a high speed to obtain . large Reynold'snumnber. Hnce air siction effects may be '±'cduced by \
turbulence full scale, compared. with a model tested in undisturbed air which in consequence might give too large a value for the 'osistance.
On the other hand a scxeened model might give too small a value.
ThIs-uncertainty is reduced in one case. If a' dynamic model i very
stable so ti
air suctioneffects on it aresmill, .thcy maybe smaller
still, full scale so that the ,aircraft is more stable,than thb model. Under these conditions resistance tests an a screened model,th air
suction effect reduced to vexAy small proportions, are 'not likely to be
mnuch in error. It follows that the logical procedure in Seaplane. Tnk testing under present conditions of test equipment and theoretical -. knowledge., is to obtain stability on a dynamic model first and then to
test a screened resistance modl. 'Resistance measurements on an unstable
model arc of very doubtful significanc.
References
No. Author Title
- Baker Tank tests on a model of tho En'ire" fly.ng boat.
Report No. B.A. 1678, 1:Iay,191 2 'G.ott Comparison of Results of Tests
- of the SitigaporcIIc Model
Hull in FiY.e Tpik R&.M No. 1785, 1937.
3 Sunderland. I Take-off tests
at high loadings. Part. of
M.A.E.E. report Nb. F/148
.kugust, 1939.
_1
0-L Clark-and Tank and Wind tunnel tsts on
Hills the Short Suniderland to
corn-pare attitudes with full scale
during take-off and lari5.ing
- R.A.E. Report o. B,1622.
July, 19401 (4732,SL45,Ae 1730)
Shaw Th effect of acceleration on
the drag of the Sunderland. hull.
M.A.E.E. report No. H/res/i 36A October, 1940
- Lerwick L7251 .Take-off tóst at
various weights. First part' of
MSA.E.E. report No. F/152. August, 1939.
V) r Attachcxi: -- i.ppend.ix -Table I V V Drg. No. 169LS 'I 1i69/Ls 14.695S ft ft 1l696S 'I ft 116971s V V Ii698 " i699S V 14700S .11 1)702S fl. " 14.703S ft 1l.4704.S
Ei,. I.
ft -1V_.. -' 2 V 3 5 6 "C 7 8 ft 9 " 10." II
IL705S " 12 Cir.clation:-V D.D.S.R1. V\ A.D./R.D.T.1 R.D.T.lc R.T.P.(T.I.B.) 2+ 1 R.T.P. - 14.5 V D.D.R.D.TS -.R.C. 36 A. D. /R. D. S. V Action copy M.& _A.E.E. Dii-ctor D.D.R.E. Library V V .Aero. (1) V T/L. T 2 JD Oott ) VInst. Det. )
¶,
V / -11-/ochnical Note No. Aero. 114.60,
r
V
Vt1 V
Title
7 Schiiidt. Scala effect in toweJ
tests
V
iith
aeipanofr0ttofl
/ units. Translation frnm V Luftfahrtforcchun. VoLl3,J-Io. 7, 1936, (2773, 3 318). V/
) I \VVJPPENDIX:
iEcasurementof airpressi.Ires in tank testing. Introduction
Followin up the work described, in nference 1, direct casurements
were rde of the low air pressurCs under the afterbody o the Bmire Boat
hocll The conditiohs peculiar to tank testinh complicate the .i-o1leri. of making the mcasuremonts. There is only a period of, say, ten seconds
at a steady speed. in the course of each run of the carriage and. bration makes it impractiahle to use any of the more usual instnments which
ar available for measuring small pressure differences. It was therefore decided. to d.elop r sLple pressure gau.ge to uit the conditions.
Di.phrarçrn Pressure Gaug
It had. been estimted that the pressure differences to be measured were of the order' of 0.3 to 0.5 inch of ater and, after a few preiminaxy
tests, a diaphraçt ras found which gave a s4b1e deflection u.dcr a
pressure of this order.. ., It was of rubber, i/6L inch thick and 'gasclamped between bloQks of wood having circular ioles, inches in diameter,
cut in them. Tests against a Chattock gauge indicated that 'the nbcr
had very little hysteresis
nd retained its calibration well, while teston the moving carriage showed. that the nathral damping of the rubber
'c-adequate to deal with carriage vibration.
Tsts on the carriage also
showed that it was necessary to, connect one side of the d.iaphrato a
static head to ohtaln a steady zero and a suitbie position for the
static head was found. Large bore tubing wa necessary to allow the
gauge to reach its steddy reading in the short time available duripg a
run.
'The deflection of the dia5hra was measured by a micrometer as
shown in fig. 10, which illustrates the gauge. Contact 1Detweenthe
plate on the diaphra and the micrometer point cqrrleted a circi4t and. lit a srall lamp.
.Exet for occasional difficulty with the
electrica contact the operation of the gauge was very oimple. and reliablc The original intention was' to multiply the deflection by a system of
iac-chapical links but in t.:.e preliminary trials the micraDeter method proved SO: simple and. reliable that it was used. in all the work. The
-calibration of th.e gauge was practically unchanged. throughout the testso /
-Theoretical considerations indicate that if a change pf pressure H pr'oduccs a small deflection d then H = D + Bd-. For the gauge. cs
used during the tets
= 0.75 a B = 23 where H is 'ex1ressed ininches of wotor and ci in inches.
3. Description af Model
The Lmplrc Boct iiodel used for the tests was fitted to a balance
frame in the noimal way for 0, resistahce model.. The hull was hollow and.
could- he quickly detached from the frame. Holes were th-illed. nonaal
to the afterbody j1aning bottom and 1/8 mci internal diameter copper
.tubOs were fittdd. to. the hclez. The tu'bes were carefully finished flush wIth thesurface and. were f'itted. internally with adapt'rs to suit 1/2 inch bor? rubber tubing. his arrangement is illustrated. in fig. 11 which also shons t. :e positions of thu three rows of hol9s in the planing
bottom. Coord.inatcs of the position are
ven in table I.
-12-Technical Note No. jero.'lL.60.
/
Technical Note No. Acre. 14.60.
Method of test
Measurements could only be made on one hole d.ifring'a run, and all
the -other holes were sealed with plasticine or denta]. wax, the seal
being carefully fath.red into the planing bottom. The rubber, tubing used
to connect the open tubto. the pressure gauge was arranged to inose as little restraint on the model as possible, arid, in the final
arrange-rnent check tests shovied that .the ri.thber tube had no ineasureabie effect on
the drag or pitching moment. To prevent drops of water from sealing the pressurc plotting hole, it was necessary to blow air through the tube while the carriage was accelerating and connect the tie to the gauge'.
wlen a' steady 'speed as reached.
Results of test
The measure presures which arc all ncgativ (suction) are giver in fig. 12. The tests were
e at a speed of 24. ft./sece (6.8 knots. full
scale) and. at dtun attitudes of 5°, 70 'and 90. These condi.tions gave
large sUctions ' Owing to the, increased speed of te air-under the
carri:ge t}ere is
static pressurä to be subtrabted' frem all these'observations. This static pressure (-0.084. inch of water' is indi.cated
TABLE I
Position of pressure plotting hales on i/i 6 scale Empire Boat model aftezbody
(A11 dimensions in inbhes)
0.95 3.00 6.96 .8,90 6 1.15 ). . .. 15 0 ) 2.4.5 11.00 24.. .2.00 ) 7 .1.10) 16 0 ) .. 2.25 1,2.90 25' 1.80) 8. 1.05 17 .0 ' 2.05 14..95 26 . 1.70 .
0.95)'
18 0 ) . 1.90 . 16,80 27 1.4.5)Technical Note No, Aero.146O.
DistanOe
aft of
point of main, step Distance from Centre line Half breadth at station 1.65 0 3.45. .. 2.90 1.60 . 0 . . 3.25 2.70 1.50 0 . 3.05 '. 2.55 . 1.4.o).' 0 2.85 2.35) 1.30) . 0 ) 2e65 2.20) . Hole No. .10. 19 2 11 . 20 3 12 21 4. . 13 22 . 5 14.DRAG. BALANCE -OLD" CARRIAGE.
DRAG BALANCE NEW CARRIAGE.
RAIL LtVEL
WATER LEVEL
POSITION OF POSITION OF
DYNAMIC MODEL RESISTANCE MODEL.
10,000 6,000 4000 2,000 1000 '000 G,000 a,000 -J w /I / I
o.
3940.
50P3' KNOTh.
B. 5G,000 LB.WT. CRNE T.NAERO f4OFIG. 2.A&6,
UN5CRENEt NEW CAR.RiME.
UNSCREENE
OLb CRRlE
\
7O
I0 O 30
4.
S0 GO 70 80SUNDERLAND
SCALE MODEL
WATER DRAGS
4/
I
1OOQ0 ,OoO LB.. oOo 400O 2,000 '14000 80O0 WATER DRAG. LB. G.000 4, 000 I000AtTITuD& - HULL DATUM
A SUNDERLAND WATER DRAGS
AT 44600 LB.
'4
AT11TtJDE HULL DATIPv1.
2- 2 GO t4G(,
F'IG.3.A &.
0. S. 46 KNOTS G9)(N0Th 12B' SUNDERLAND WATER DRAGS AT 56,000 LB..
15-4 KNOTS
.
KNOTS.
6OO KNO
to,000
B000
;,000
4,000
2,000.
A. SUNDERLAND WATER. DRAG AT
VARIOUS. WEIGHTS
100 80
60
40 20 e .H
500O L;
1I60,000 L
. 44,G00 La.V
/1 ...H
eo 30 40 50 0 70 - KNOTS MO.B.
SUNDERLAND - TAKE -OFF
TIMES AT VARIOUS. WEIGHTS
J ERO. %4o.
FIG. 4A&B..
DEL. \ tILL SCALE 30,000 40,000 50000 WE)CMTLB.B4O00 / 80 20 026000 28.000 SCRENtD UNSCREE WED a 33,000 LE. 30000 LB. 8,000 LB. 30 40 50 0 10 80 SPEED .- KNQT.
LEICK WATER.
bRAG.
AT VARIOUS WEIGHTS.
0
MODEL
FULL SCALE
300.00' 3Z.000 . 34,000
WEIGHT- LB.
LERWCK TAKE OFF TIMES AT
VAR1OUS WEIGHTS.
-TN. AR
.FIG:s A, B.
000 000 p00 I,o0 r01000 oo ooO
UNSCREENED MODELS.
TN. AtRQ. tG0FIG.6A&B..
-J SU4RLAND HULL ON LERCK. / / / / I,/
I I I LER WICK.SCREENED MODELS.
90 I /F'
I \ . . . ----.---.--._N
-
----..-,.SLJNOERLANO NON LRWICK. .'-.--lULL.1
/
I I 1. LRWICK . -. .,
/
N 10: 50 60 80 0SEO-KNOTS-COMPARISON OF LERWICK AND SUNDERLAND.
WEIGHT 28,000LB.
I-20 .O 40 50 70. 60
-J B 4 8000 000 WAT ER DRAG LB 4000 0 z000 SPEED-KNOTS. SPEED KNOTS.
EFFECT OF SPOILER RIDGES AND SCREEN Ot'ql TRIMMED TAKE - OFF DRAGS.
TN.AEkO 1460
FIG7
EMPIRE BOAT (!/i6 SCALE MODEL) 40 500 L B.
TYPICAL ASSUMED TAKE - OFF ATTITUDE. 10 0 40 SO 60 70 Si . . ,/UWSCREEP MODEL -WITH SPO RIDGES.
/
LER N POINTS +'ARE MODEL SCREENED . ) 10 0 40 50 60 70 84000 5,000 4000 3000 LB 000 000 .3000 LB. oo0 THRUST. .30 40 0 0 70 SPEtD-KMO7S.. ALR ORACi. AIR DRAC. TOTAL RA TOTAL DRPC1. SCREENED.
/
I -,ScRN!fl.
SCRNb.
FLAPS AT435° UttISCREENEO. 10 . 20 30 40 0 0. 80 .SPED- KNQT..BARRACUDA
- TWIN
FLOATS
SCALE MODEL
12000 LB WEIGHT
TN. AERO. %4.
20o o t500 boo so3 e000 t500 1000 500 T.N.AERO. 14G0.
FIG9
30
'7 <jKJ w 'I) 3 L p.-Ui a I-20° C 80/
/ AT7ITU1E /----"/1SCREENED
flRAQ. UNSC.REENEQ QfAC. 1 / I ATTITUDE.S.---
-ii
/
UNSCREENE SCREENED DRPj.SPITFIRE ON FLOATS
CSCALE MODEL
7650 LB
0 10 O 30 40 50 70 SPEED-S ).NOr5 0 20 30 40 50 GO 70 - KNOT5.Lu-U.
-. It \,I
1 .-.----.-. it; I-\s'
---I -I. -L -L_ '64 RUBBER. nIAPwAr:KA TO B AT TERYAPPROX FULL. SCALE..
SIGNAL LIGHT
/
MICROMETER CONTACT ADJUSTMENT.
PERSPEX COVER PLATE
DIAGRAM OF DIAPHRAGM PRESSURE G:UAGE
CON!
/
TO STATIC HEPD.T.N.AERO. 14G0
/
FIG. II.
I T0.
SECTION OF EMPIRE BOAT MODEL SHOWING
PRESSURE PLOTTING TUBES.
20 21 22 23 z4 25 2G I p I p p + MAIN S1!P 15 1? a 12 13 p 10 _I_ REAR SIC
LOCATION OF FRESSURE
PLOTTING TUBES ON
ArIERB0DY EOTTOM.
19
PRESSURE kmiu4(: (y
AFTERBOby
SPEED
4
SEC..
TN. AERG.1460.