Model Hovercraft Technical Suggestion.

Matching a centrifugal fan to a chosen model size.

Brian Wise 22/02/2003 Here's m method for lift fan testing. ! don't claim an scientific so"histication for it# it's $ust a starting "oint...... %ll ou do is &model& the model. The main idea is to e'haust the fan into a fi'ed volume (hich re"resents the s)irt volume and the airs"ace under the model. The outlet of the fi'ed volume re"resents the &hoverga"&# the area *et(een the edge of the s)irt and the ground through (hich the air esca"es. % load "laced on to" of the *ag re"resents the re+uired cushion "ressure. Success is $udged * the a*ilit of the s stem to su""ort the load. This needs to *e +uic) and eas to do# so there's a com"romise in that the &model& is a tu*e and inevita*l (on't accuratel re"resent the airflo( in the real model. Ho(ever# it (il, *e "ossi*le to esta*lish a (or)ing figure for cushion "ressure and airflo( for a "articular fan/model com*ination. B using an estimated s)irt volume it is also "ossi*le to o*tain an idea of ho( +uic)l the s)irt could res"ond to distur*ances. -.hoto0,/ Ste" 0,1 2stimate the s)irt volume and the airs"ace volume under the model. This doesn't need to *e su"er accurate 3 a +uic) method is to calculate4 hardstructure length ' (idth ' side*od height ' 2 0assuming an 5/25 *ag s)irt1 and ignore the volume of the hull under the side*od line. 2'am"le4 for Mar) .orter's 2000T67 design this gives ,08cm ' 90cm ' :cm ' 2 ; :<.2 litres % more "recise calculation gives4 =uter s)irt volume4 ,08cm > 90cm ' 0"i1r?2 (here r ; :cm ; 2<.0, litres %irs"ace under4 @<cm ' 22cm ' :cm ; ,2.A litres !nner s)irt and air s"ace 0side*odies14 2 ' ,<cm ' ,08cm ' :cm ; 20.:: litres ' 2 02sides1 ; <,.< litres Total ; ,2.A > <,.< > 2<.0, ; :@.3 litres Ste" 021 Ma)e a tu*e from suita*le "ol thene *ags to re"resent the volume from 0,1 Bse a radius for the tu*e of a*out dou*le the s)irt radius# other(ise the tu*e (ill *e too longC 6omestic freezer *ags 303mm ' <30mm (ill *e o). Dut the sealed ends off and stic) end to end (ith 6uc)/Eaffer ta"e. Feave a *ag to *ag overla" of ,0mm. Ste"031 Ta"e the fan unit to one end of the *ag and a fi'ture of suita*le cross sectional area re"resenting the hoverga" to the other end. Bse a ogurt "ot rim or short card*oard tu*e# "lastic "i"e etc. !f necessar use t(o or more tu*es of the same diameter to give the cross sectional area re+uired. 2'am"le4 for 2000T67 the hoverga" area could *e4 02 ' cushion length "lus 2 ' cushion (idth1 ' height of s)irt edge a*ove the ground ; air esca"e area 00,08cm ' 21>090cm ' 211 ' 0.29cm ; 02,2cm > ,00cm1 ' 0.29cm ; :@cm?2 % suita*le tu*e is then :@/0"i1 ; r?2# r ; <.A@cm 09cm1. Bse a tu*e ,0cm diameter. -.hoto02/ Ste" 0<1 Dhoose a suita*le "iece of (ood to simulate the li)el loading. 2'am"le4 38& ' 2& gives 3 ' 0.,88: ; 0.9 s+ ft Weight of (ood ; 0.8:l*s# therefore loading ; 0.8:/0.9 ; ,.3< l*s/s+ ft Ste"091 .lace (ood on "ol *ag tu*e# run motor u"# "refera*l using a s"eed controller. With gear*o'es this is essential to avoid *rea)ing a *elt or stri""ing gearteeth. Fift Gan Test 5ig ,

Ste" 081 Will the fan su""ort the (eightHH !f it does# the &hull& should rise from the floor and &hang& in free s"ace (ith no visi*le means of su""ort 0other than the "ol *ag1. Which ! find +uite fascinating# es"eciall if clear "ol *ags are usedCC -.hoto03/ Ste" 0:1 Iar the outlet diameter# fan r"m and cushion loading to esta*lish a (or)ing "oint. Dollect the data# "onder the results# *uild the modelCC Jotes4 The "ressure in the *ag (ill *e ver lo(# for ,.9 l*s/s+ft this (ill onl *e :2 "ascals# 0less than 0.3 inches of (ater1 so conventional methods for "ressure measurement 0eg an inclined slo"e manometer1 (ill *e difficult to use 0*ut not im"ossi*le1. Donse+uentl # the *est (a of assessing the air "ressure is to var the *ag contact area of the (ooden load and/or to add/remove (eights to see (hat ha""ens. % variation is to "artiall *alance the (eight of the (ood (ith a s"ring *alance to var the loading# the difference *et(een the (ood and the *alance gives the effective (eight of the (ood. -.hoto0</ This "art is +uite su*$ective 0de"ends on our "oint of vie(1. ! have classed the "ressure in the *ag at a "articular loading as follo(s4 Soft4 the *ag is $ust a*le to su""ort the load# sides of the *ag are curling round the edges of the load. The *ag is soft and sagg # it (on't maintain a circular or near circular cross section# *ag "ressure is too lo(. =K4 the *ag su""orts the load# *ag cross section is such that the load edges are still in contact 0$ust1 (ith the *ag. The onl flat "art of the *ag is directl under the load. Girm4 Bag su""orts the load# edges of the load are off the *ag# load tends to roll off the *ag. %t this "oint the *ag "ressure (ill *e o*viousl too high. 6iscussion4 %ll the 0su*$ectiveC1 *ag "ressure results (ill relate to a "articular set u" (ith a "articular air outlet area. !f the results sho( a satisfactor *ag "ressure (ith our fan at a "articular 5.M then it (ill *e li)el to (or) on the model. !t 's also advisa*le to allo( for losses in the model's airducting due to tur*ulence# so add a*out 20L more flo( to the calculated figure. !n most cases this can *e o*tained * increasing the fan s"eed. !f "o(er consum"tion is too high it ma *e *etter to increase the diameter of the fan# change the motor or gear*o'# etc# etc. !f the tests had *een done on a model then the *ag "ressure classifications (ould *e irrelevant# the (ouldn't *e noticed *ecause the airflo(/*ag "ressure relationshi" (ould have *een automaticall translated into a hoverga" change# ta)ing advantage of the inherent automatic ad$ustment in the s stem. The +uestion for an electric "o(ered s stem is4 ho( much hoverga" do ou needH 0it (ill change de"ending on the roughness of the o"erating surface1# *ecause an increase in the hoverga" (ill re"resent an increase in airflo( and an unnecessar drain on the *atteries and shorten the run time. ! tend to (or) on 2.9mm or 0.,& for the hoverga"# this a""ears to give reasona*le results for airflo( and "o(er consum"tion. Donclusion4 Wh *otherH !t can *e e'"ensive to *uild a ne( model and it's disa""ointing to find out that more lift "o(er ma *e needed. %t least (ith this method# time and materials are not (asted. %ssuming that a ducted fan unit is *eing tested and all the "arts are to hand# then a*out an hour should suffice. Gor standard "ro"ellor testing a duct of some sort needs to *e made# "ossi*l using st rofoam etc. Gor a centrifugal fan it could *e mounted *et(een t(o "arallel "lates. %ha ou sa C We might as (ell *uild the modelC Well es# (h not# it (ill still ta)e timeC 0There's all that stitching to doC1 Fift Gan Test 5ig 2

Wor)ed e'am"le4 .erformance curves for small centrifugal fans li)el to *e suita*le for model hovercraft use are availa*le from several manufacturer's (e*sites. ! hesitate to recommend an "articular fan as (e are li)el to use it outside of its intended design "arameters# instantl # * modif ing it to fit to a model electric motor and conse+uentl incurr criticism. Gor fan modifications# ou are on our o(n CC Fet's tr to match a t "ical 9.29M diameter fan to Mar) .orter's Eriffon 2000T67 model. 0!t's the most "o"ular model around and several s"ecifications are availa*le1. Model (eight (ill *e 2.:0Kg 09l*s ,9ozs1# on a cushion area of ,.08 ' 0.9m ; 0.93m?2# gives a cushion "ressure of 9.0A)g/m?2# or 90 "ascals. 0 , .ascal ; ,J/m?2# 9.0A)g/m?2 ' A.@,m/s ; 90 "ascals 1 2+uivalent to ,.0<<l*s/s+ft. 0Dhoice of model (eight is a *it o"timistic 0C1 *ut is intended to match the fan's "erformance gra"h1 Grom the "u*lished data for the Nt "icalM mains "o(ered fan# the air free flo( is 280m?3/hour 0,92 DGM1# and at 90 "ascals "ressure# the air flo( ; 2,0m?3/hour 0,23DGM1. Ho( much airflo( is needed for the modelHH Grom information availa*le on the THDD (e*site and the Fift Dalculator at Hoverha()s.com# (e can "erform some sim"le maths. 0!ncidentall # tr the Hoverha()s lift calculator (ith the dimensions for Mar)'s Eriffon and ou (ill find that it (or)s for model dimensions of ,.08m * 0.9m# (eight 9)g 0,,.023l*s1 giving a cushion "ressure of A2.9 "ascals 0,.A3l*s/s+ft1. !t (ill allo( ou to go do(n to a*out @: .a 0,.@,:l*s/s+ft1 There is a*out a ,:L addition to the airflo( figures# "resuma*l to com"ensate for losses1 Theoreticall # The esca"e velocit 0Ie1 of the air in m/s is4 Ie ; s+ root02 ' .c/"1 (here Ie is the esca"e velocit of air in metres/second# .c is the cushion "ressure in "ascals# " is the densit of air in )g/m?3 ie4 Ie ; s+ root 02 ' 90/,.221 ; A.09< m/s The hoverga" area ; length of cushion ground contact ' hoverheight ie4 02 ' ,.08m1 > 02 ' 0.9m1 ' 0.0029m ; 3.,2m ' 0.0029m ; 0.00:@m?2# 0:@ s+cm1 gives an outlet diameter of4 :@/"i ; r?2# r ; <.A@cm# 0use a ,0cm diam "i"e.1 The volume of air re+uired is4 hoverga" area ' esca"e velocit # ie4 0.00:@m?2 ' A.09<m/s ; 0.0:082 m?3/s or ,<A DGM. Grom the "erformance gra"h (e find that the airflo( at 90 "ascals is too lo(# it's onl 2,0m?3/hour or ,23 DGM. =ur theoretical hoverga" (ill *ecome instead4 2.089 mm. !t's still set to 2.9mm as (e haven't changed the outlet area# so (e could instead e'"ect the cushion "ressure to *e less to com"ensate. Bsing the test rig (e can test this arrangement. The fan can onl run at one s"eed 02@00r"m1 so the load and the outlet area are the varia*les that (e can change. With a ,0cm diameter outlet# the *ag su""orted a load of ,9.2,)g/m?2 ; 3.,29 l*s/s+ft and (as definitel too soft# 0ie the load (as too much1# indicating "oor "erformance. Bsing the s"ring *alance to su""ort a*out Aozs of the load# this gave an effective (eight of ,: 3 A ; @ozs giving :.,9 )g/m?2 or ,.<:l*s/s+ft. *ased on the criteria for the *ag sha"e indicating cushion "ressure. !n "ractice# it (as uncertain (here the actual load "oint (as# 0from estimating the *ag sha"e1. !t (as eas to see if the loading (as too much# not so eas to see if it (as too little or near the re+uired (or)ing "oint. Fift Gan Test 5ig 3

The s"ring *alance method can give a first a""ro'imation# then use this loading value (ith no s"ring *alance su""ort to get a closer result. !nitial da**ling suggests that an a""ro'imate 90L overload can *e tolerated and not *e noticed in the *ag sha"e. The theoretical change in the hover height due to the lo(er airflo( (ill *e so small that it (on't *e noticed# unless the s)irt# 0"articularl a *ag s)irt1 tends to stic) on (et ground in our local car "ar)C !f ou are ha"" (ith the cushion sha"e then that's =KCC 0Ho(ever# (ords of (isdom heard at a Hovercraft Societ meeting (ere4 & ou can never have too much lift air&1 Jote that (ith this e'am"le# there is unli)el to *e an reserve lift *ecause it has *een Nused u"M * the "arameters set for the test. =*viousl (ith a "ro"er dc motor setu" (ith 2SD# reserve lift is availa*le * increasing the fan s"eed. The same a""roach can *e used to test ducted fans and "ro"ellor drives. Gor these# it can *e e'"ected that the s stem (ill *e more sensitive to changes in loading *ecause the "erformance curve for an a'ial fan/"ro"ellor s stem (ill *e much flatter than for a centrifugal fan s stem. Farger changes in fan r"m (ill *e needed to com"ensate for small loading changes. 2nhancements4 Gor a &"ro"er& model setu" 0com"ared to a mains "o(ered fan1 it (ould *e useful to also measure fan r"m# motor current and run time to allo( motor/fan com*inations to *e com"ared. Bse an inclined slo"e manometer for more accurate cushion "ressure readings. Gootnote4 ! deli*eratel left out an reference to measuring the air velocit from the *ag outlet for t(o reasons# first the cost of an air s"eed measuring device of an sort is +uite high relative to the cost of the "ol *ags and stic) ta"e and "ro*a*l (ould onl confuse things *ecause# secondl # ! have alread tried it (ith a Kestrel (ind s"eed meterC The *ag outlet air velocit is not uniform across the outlet area# it's at a ma'imum in the centre and dro"s off at the edges. %lso the size of the measuring device can influence the readings * *loc)ing "art of the e'it area. Suffice to sa that des"ite the "ractical "ro*lems of o*taining a reading# the air velocities usuall agreed (ith the calculated velocities. Donse+uentl ! left out air velocit measurement# instead to rel on the *ag "ressure sha"e to indicate success or failure. !f the *ag "ressure is a*out right then the airflo( must at least e+ual the calculated figure# if not# $ust turn u" the s"eed control for more air flo(CC !t ma seem that ever thing has *een grossl oversim"lified and things left out# for e'am"le no mention of discharge coefficients# no formulae to relate "o(er consum"tion to airflo( or "ressure. Gor "ractical modelling (e are limited * (hat is availa*le commerciall # most of us lac) the facilities to start from scratch# so most "arameters have alread *een chosen for us# it $ust remains to match the fan and "o(er s stem to a chosen model size# or conversel # (or) *ac)(ards and find a model size for a "articular fan s stem. .S. 6isclaimerC %n modifications to commerciall availa*le fans should *e a""roached (ith caution. .articular attention should *e directed to4 0,1 "ro"er# secure mounting of the fan# 021 *alancing in 2 "lanes to minimize vi*ration# 031 for safet # avoid using metal im"ellers# "lastic ones are "referred. %ll comments a""l to models onl CC Gull size design is a""roached in a far more so"histicated (a CC %"ologies to all )no(ledgea*le "arties# (ho * no(# after reading this# are either laughing u" their sleeves or have *een violentl sic)CC 5eferences4 THDD (e*site# hovercraft "rinci"les Hoverha()s.com# lift calculator. Eriffon 2000T67 model design at (((.model3hovercraft.com 5egards# Brian Wise. Do" right B.H.Wise 2003. Girst draft4 28/02/2002. 5e(rite and additions4 22/02/2003. 2dited ,@/,,/2003# version 3.,. 5e3edit 29/,,/2003 and ,,/,2/2003# version 3.2

Fift Gan Test 5ig

<

.hoto ,. Eeneral vie(# fan unit at front# air outlet at far end# (ooden load on to" of tu*e.

.hoto 2. %ir outlet using ,0cm diameter "lastic "i"e# simulates air ga".

.hoto 3. Wooden load# simulates cushion "ressure loading.

.hoto <. S"ring *alance used to reduce and measure loading changes.

Fift Gan Test 5ig