· Form ApprovedREPORT DOCUMENTATION PAGE OMB No. 0704-0188

The public reporting burden for this collection of information is estimated to average 1 holX p8f response, including the tirTl8 for reviewing instructions, searching existing data so~ces. gathering and ma;,ltaining the data needed, and completing and reviewing the colJeet;on of information. Send comment:; reg_ding this burden estim~ or any c::>ther aspect of thIS collectIOn of information, including suggestions for reducing the burden. to Department of Defense. Washington Helldquarter~ Services, Directorate for Informatlon Operatrons and Report:; (0704-0188). 1215 Jeffer:oon Davi:: Highway, Suite 1204, Arlington, VA 22202-4302. Respondent::: !:houJd be aWare that notwitt'lstaru:ting any other provision of law. no person shall' be subject to any penalty for failing to comply with a collection 01 information if it does not display a currently valid OMS control number.

PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS. 3. DATES COVERED (From - To)1. REPORT DATE (DD-MM- YYYY) 2 REPORT TYPE

1 . June 2009 - January 201020-01-2010 Final Technical Report Sa. CONTRACT NUMBER 4. TITLE ANO SUBTITLE

W31P4Q-09-C-0548 Hazardous Particle Detection via Unmanned Air Vehicles: Optimal Placement 5b. GRANT NUMBER of Sensors in Forward Flight

5c. PROGRAM ELEMENT NUMBER

5d. PROJECT NUMBER 6. AUTHORIS)

Drs. Peter S. Bernard and Jacob Krispin

5e. TASK NUMBER

Sf. WORK UNIT NUMBER

8. PERFORMING ORGANIZATION REPORT NUMBER

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESSlES)

Vorcat, Inc TR-IO-OOOI14 Freas Court

North Potomac, Maryland 20878

10. SPONSOR/MONITOR'S ACRONYMlS) 9. SPONSORING/MONITORING AGENCY NAMElS) AND ADDRESSlES)

Defense Advanced Research Projects Agency DARPA Suite 120 3701 Fairfax Drive, Arlington, VA 22203-1700 11. SPONSOR/MONITOR'S REPORT

NUMBEB(SI

12. DISTRIBUTION/AVAILABILITY STATEMENT

Approved for public release, distribution u.nlimited

13. SUPPLEMENTARY NOTES

14. ABSTRACT

This project explores means of enhancing the efficiency ofbio/chemlnuclear hazard detection in the atmosphere by WlD1anned air vehicles (UAV). Specifically, the study seeks to discover if and where the placement of sensors on UAV's can maximize the volume of space that is brought into contact with the sensor. A hybrid gridfree/fmite volume nwnerical flow simulation methodology that is adept at modeling complex flow scenarios provides the basis for the analysis. This study focuses on the use of an unmanned helicopter in collecting particle data. Clear evidence is found that the judicious placement of sensors on the UAV can profoundly affect the efficiency with which the atmosphere can be surveyed for hazards. In particular. for the geometry considered here, sensor placement in the tail section of a helicopter in forward flight is found to enable sampling particles that originate from relatively widely separated forward positions. Future work should be devoted to optimizing sensor placement for particular geometries under a range of flight conditions.

15. SUBJECT TERMS

16. SECURITY CLASSIFICATION OF:

a. REPORT b.ABSTRACT c. THIS PAGE

17. LIMITATION OF ABSTRACT

18. NUMBER OF PAGES

"19a. NAME OF RESPONSIBLE PERSON

Jacob Krispin ......~~-_.~-......._-_..~

U U U UU 19b. TELEPHONE NUM13ER (/nclude Bfea code)

301-762-5553

Standard Form 298 IRev. 8/98) Prescribed by ANSI Std. Z39.18

.. .. . L\L\ Advanced Development for Defense Science and

Technology

January 20, 2010

Sponsored by

Defense Advanced Research Projects Agency (DOD) (Controlling DARPA Office)

Under ARPA Order AXll-OO

Issued by U.S. Army Aviation and Missile Command

Contract No, W3lP4Q-09-C-0548

Prepared By: Dr. Jacob Krispin, PI

Vorcat, Inc. 14 Freas Court

North Potomac, MD 20878 Tel. (301) 762-5553

email: jacob@vorcat.com Home Page: www.vorcat.com

SBIR Phase I Final Report

Title: Hazardous Part'c e Detect'on via Unmanned Air Vehicles: Optimal Placement of Sensors in Forward Flight

Period Covered: June 26, 2009 - January 20, 2010

Approved for public release; distribution unlimited

D SCL tIER: The views and 'ond .ons '0 tained in thi document are those of the authors and should not be interpreted as representing the

official policies, either express or implied, of the Defense Advanced Research Projects Agencv or the U.S. Government.

A ract

This proje·t explore means of enhancing the efficieuey of hio/chem/nuclear hazard dp.­t.ect.ion in the atmosphere by Ilnmanned air vehicle:; ( AV's). Sp cifically. the st.lld 'seeks 10 ',c r if and wh re t lac of sen 0" on AV me ximiz the voLn of space t.hat is brought into cont.a't with tbe sen, or. A hybrid gridfree/finjt volume numerical flow simulation ill odology that is ad a modeling mplex f10\''' scenar' s provines th hasis for th0 Ana ,'sis. This st.lId.y focHs .S on the lISe of aD llnmanl1 d heli­cop 'n co cti par iet cl . Cl FIT \'d nee is fmmel tl a tho . IdiciOll. lacement of s nsOJ' on he UAV prof undly eet th efficiency with \vhieh the atmo here an be 1'llrv . ed for hazards. partie la , £ r the cas S Dsid red t hU8 f r. se sor placement in th tail sec i n of th h licopt . in forw' d flight is found to enable sarnpling parti 1 s that ori 'nat from relatively win 'I, r s parat d forward positions. Futnre work should be • devot d t.o optimizing sensor plac ment for particular geometries under a mnrre of flight conditions.

Summary

Pra.ct.ieal scenarios wer explored for m:'\ximizing the efficiency wit.h which unmann , Ftir vehicles. specific811~r beli opters, can be llsed to sample the atmospher for rh m/bio/nllcleM haz8rds. Simnlation of part.ide flows nCOlwt.erecl by a, helicopt r in forwATcl flight ,ver carried out usin the VorCat code. A s'n ariv effecti e means of accomplishinrr the proj ct o<1,1s was ide )tificcl: pFtftirles 8rriving t the r e1.r il "ti n of th helieopt . air frame are fo ill to have a tendency to ori inate in widely separated forw 1 po_ i­tions. Plaement of P' ii 1 sensors in this locat,'on will be most efficient in samplinp' th" atmos here. The methodol qv cl v loped here can be extendeo t.o provide detailed rec­

mmennat.ions for sensor placement. in s ecific helicopter O' metries ftnd flow conditions.

1 trod ctio .

The detection of hazardous airborne chem/bio/nnclear paTticuJates via sensors carried by unmanned air vehicles (DAV's) provides a potentially safe and effe tiv mans of gaining vAlml.ble t.ime for mit.igatiJl or mini izing th cons queue s of such bazftrds t,o civiliftn and mili ary populations. n important pFtf of tl,is tecbn logv is optimizino- tbe efficiency with which the fttmosph r . n be n d~ a oal t.hat is synonymous with maximizing the reg'on of the atm sphere from which )articJ. nllp'ht arri" at the _ boaro s nsors. This proj ct con"idp OR how th complex. t..lll' ul vortical flows prodllced by DAV's can be harnessed for this pmpose.

The eEre tivene of current point me airborn testi f r baz nous parti 1 s using V's is limiteo by the n rl to directly int rs ct t h p8Tticles alono- the fti ht path

[1, 12, 191. Thjs laeJ( of effic:ienc an latc~s int.o extended flight tlmes and th potentiality for mission failure in which danCFerous pa.rticles are overlooked [15]. While it is po. sibl to force increased airflow through sensors, (e. O". by the use of fans [20n it is not. evident that this necessaril in T as , the R arch volume in tll snrroundinf'" ~tir ,pace. The fOC115 of this proj ct is in finding tr tecries for creating the laro"est possible sampling footprint of t.he UAV as i ' flies thus enhanc:in Jfi 'ienc, a.n ac lITacy in the s arch for da.ngerous particles,

The feasibility of this study is sng 'e~'3ted by the tencienc)r of UAV Row fields [14. 18} to includ a range of om 1lex flow phenom oa that may include vort.i 'es associa'ed wit]l separated flow and vehicle manenverinfT. vortex bre ·clown. shear layer instabilities and other eff ds. VOlt x motioILS axe also llltrinsir to flow control efforts inducting mkro­electro mechanical (MEMS devices. Flow fields containing vorti es and turbulence are note'worthy for their ahilt y to promote dispersion ann hy so doi ma brin IT particlesIT

'" ithin range of the in ake f t.h airh me sensor that would therwi he ov rlooke . While t.mhulence may s ronlJ'ly promote part.icle dispersion; he extent and nat,me of the effect vaxies dep cling on the volume and wicrht of th arb ·1 s. som s. he effe t of turhulence can b quite specific [22, 1fi, 11] and pot€nt.iall)r useful in data

colle tian. Thlls. an im )ort,ant ad Etionc I cOlLsideration in h roj ,is to dis Tn what kinds of particles ar likely to be most advantageollsly bronght to the sensor by specifi strategi s £ r air c L1ection.

For the o'oals of the project to be attainable. a numeri aJ scheme is reqllir thFtt can on the one hand ac urat ly repr nt th mp x n n-st dy dy' motion a1:S iat dIT

'with UAV's and on the other cap UTe the essential dyna"ill ics of how a.irborne par icles of differ t sizps d weigh ,s diRperse within h€' turhu.l nt flow e vironment . Sin llnstea..(hne,'s is an essential Ftspect of the p ysics. it is natural to use a Large Eddy Simulation (LES) method to represent the flow [21 in his offers he best pro p ct for rpduced modelin at an aHorclab1 cost. Among; LES a.lternatives. the gridfree vortex fi1 nt cherne \TS d by VarCat , Inc. is uniql e in it biHt 0 captur· the phy ical dispersion of particulates without the unph,vsi al effect of imposed or nnwanted sm I s .. iliff 1 ion t ha.t is the haUm rk of tradiional i l-h88 d LES schemes. The availa 'lity of this Dew met.hodology is an important. factor in helieving that t.he goals of t.he project are f~ ih ..

The use of gTicifree vortical el ments provides VorCat with a direct representation of the pssential phrsi 1 properties of tnr 1llent :flow. This promo AS 8 inifica:l gain in numerical effiden y sin e t.he vortices that form the primary strnctme of turbulence are modeled direct.]y \vithont the int TV ntion of a D1lIDerica.1 mesh. 10reav r. it is the sence of a mesh that removes the primary source of num ricaJ diffusion in 'omputational fluid dynamic,;." (CFD) Rchem s. A Dumber of vaIi < i II stndieR [3,4. ',6, 8} have demon­strated the capabilities f the VorCat approa.ch and SUO"rrest that it is well snit] for tb . pr p ed work. For exampl ; the VorCat Rch illP hFts the prop rty that it is appli d the sa.me to a.ll flow sitnations: it does not require the, electio.n of modeling pa.ramet rs

4

whose precise va-Ines determine whet.her or not important physical trends :=:tre ca,ptnred.

2 Methods, Assumptions, and Procedures

This st.ndy is ba.<;en on the opportuniti s for ~.md)Tsis And simuIA,t.ion of compl x flows provided by the VorCat. Inc. hvbrid gridfre Jfinite volnme flow solver. An overview of the technolo ii:' provided here inclndin,O" a di;~cm; i n of how it is ap lied to the particlllAI'T

Bow of lit T s in tb pres nt pro' ct.

2.1 Overview of the VorCat Methodology.

Vortex Tubes. Straight va t x tube, forming til en ar grjdfree I nent u in VorCat. The ith tube is identified via its end points xL x; and circulation rio Vortex • tretchinrr and reorientaJion a! ontained in t.he overnin' v rticit.y transport eqnation are accommodated hy convecting the tnbe end points according to the 10 A,l velocity fielrl. Tubes tha s r t,cb hey nd a maxjmlill1 length hare suhdivided. The circnla ion of th tnbes is a.ssum d t,o remain constant in time owing to Kelvin's theorem: a reaR nable a pro.ximation for biO"h Reynolds munber flow aW8:V from honndaries. Stretching and folding of the vortex tuhes cillTi s energy to srnalJ clissipAt,iv s ales. Small vort.ex loops that naturally form are removed [4, 9, 1OJ. t.hns providing a model of small scale 10 1 energy eUssipa,tion thi'l.t. is Don-diffnsive an sspntia,l for numeric;',!.! efficiency.

Finite Volume Scheme. Since solid honndar'es ar h' n \V vorticity is common], . proollced in tnrbnlent flow fields, it is vitAlly impor ant to a Cllfat Iy simulate this pI' C ss as PaJ"t. of the VorCat. algorit,hm. Conseqnent.ly, VorCA,t mploys a hip-h resolution finite volnme seh me appli to t.he c mplet.e 3D viscous vorticity eqnation t,o compnte the flow ad' acent to walls. Solid. 11rfa.ces are represented throngh triangnlarizations, ont of which a t.hin mesh of t,riAn~mlflx prisms is grown 1y er ding er eli ulars the nodal points. e nnmber of layeT~ is usuallv taken to be 11 witb t,he first Lay of half­thickness. The prisms ideall bFt" pect. ratio of 10. The thj kness of the m sh region i. esigned to encompass the high vorticity and it gradien s that are pI' ducen next to solid \valls via viseosi ,yo Thi r gion g n rally onstitntes the vis OllS snhlayer and tin

imm eliately a.(:l'acent part of the tmbulent h ilmd::uv la\' '. Tb no-slip audition is llsed to eva,luat.p vort.icity at the surface. Vo.rtic.ity h t conve and cliffllse to the top sheet layer is turned into new vort.ex tubes. Details of the finit volume scheme may be found in recent articles [7, 8].

Velocity Field Th vorticit. residing in the t.ube" an in trianglliar prism ntrihutes to the velocity field as a sum of tbe eontribllt.ions from inoividual elements fl ' 'oTding to the Biot-Savart law [21], For t.h i h tube ent. the smoothed velocity field induced

,5

a.t a point x may be computed from

(1)

where f i is the circulation, r;. = Xi - x, r =1 r I. ¢(T) = 1 - (1 - ~r3) e-r~ is a high nrd r moot.hing flillct.ion, Si is the axial vector along the egment and (T is a scaling param tel'. In the ca. e of sheet.s, the velocity field due to the ith sheet is determined hy inteO"ratin the Biot-S< WITt. kernel over the prism volume. These exact forum las are lISed in local calclliations. and the prisms e re arded as additi nal tubes for more distant contriblltions. To enforce inflow and outflow bonuoary c lditions or th non-penetration b01md y condition at solid waUs, R id-fr e potential fl W Rolllt.ion 10 O"in rr to a surface source distribution is added to the velocity clue to tubes and prisms. An advanced, parallel implementation of the Fast Multipole Method (FMM) [13] is used to compute the velocity produced by the N vortices and prisms in O(N) operationR.

Validation Studies. A variety of b ic flow have been s udied usin0" VorCat in the interest of validating the technology. Computation of an isotropic turbulent field [4J es­tablished the capabilities of the method in capturing a ranCTe of fundamental flow ~ tatu tics inclllcling spedra" correlation and stmdnre functions. A snhseqnen compntat.ion nf the spatially developing mixino- layer [;::] has provided the most comprehensive simulation of this important flow t.o date. A significant ac:complishm nt of this work is providino' an a.c· 'ma.t. portr-. al of t,be flow from laminar inlet. through the development of vorti I strllctmes in transition to a. fully turbulent self-similar reo-ion. Compllted statistics and RtI'll tnral aRpects of th mixing lay r match previolls pbysic I eA'])eriments.

Similar accuracy has been achieved in the simulation of a co-flowing jet [6] and most recently in an investigation of the turbulent boundar layer [7, 8}. The latter is notable for captnring the complete transition inclllding Blasius and tmhulent log law velocity pI' file. Mon~ov r. t.h nnjque vi -w of the houndary 1, 'er struet e offered by vort x filA-ment.s. tha.t give information abollt the orienta ion and organization of the vorti ity field. has provided significa.nt Ilew in;;ight.s int.o how tbe bOllndary laver is or anized and evolves.

2.2 Helicopter Flow Including Particulates.

After discussions wit.h Boeing, In '. tb decision was m to limit the t.vpe of llnmann d air vehicleR consiclered in the Pha$€ I st.udy to unmanned helicopters since these offer t.he

e<i.test npportllnity for baruessino- th roperti R of h ]nca,} flovv field to RA:complish the goa 1s of the st.ndy. In this proof-of-concept study a n ric heIi 'opter sha.pe is usedO'

to T pr Rent t.he air fn:u:ne in tb alculations 8,od the r t r is modeled by introdll inO" new vorticity from the blade tips as th y rot8,te. An 8,rbitrarv nnmher of blades CI'lD

be p cified aH well as a contil11101LS ring of vorticity in the limit of an infinit 111 Imber

6

-0 'r ::l -0.1 ~

~0:1 0.2

:~f 0.•1 , ,

-1 -0.8 -iLO -0.4 -0.2 y'

Figure 1: Side view of the problem setup. Out line of 0- neric helicopter shape is hlue. rotor pI e and xis is r d, flight direction is green 8.n planes of particles rele::t.sed every 40 time steps (a.pproximatelv 0,2 time units) upstream of the helicopter are cyan.

of blade, as in the actuator disk model. Fi me 1 shows the g n r I • tup as viewed from the side. Th <Treen horiz01 tal line incli :8.tes the direct.ion of flight. the rotor disk is r d an circillar planes of particles l1pstream of the helicopter are ·yan. Cal ll.ations are perfoffil d in 8 referencB f arne fixed to t.he helicopter with the planes of particles released periodically every 40 time t ps of the computation. Th passage of the particle, through the flow fi lei prodllced by the rotor and helicopter body sh8p is computed 8no analvzed to determine an optimal placement of the particle s nsor.

The velocity field is sC::lled by the nominal velocit. jllSt d wDstream of the rotor so this is nnity in the comput<'ttional problem. For the r suIts dis llssed here the heli opter is tilted at an angle of 17.7° from the horizontal with tb conseqll nee that the helicopter experi nee", a flow of 0.2 normal to the rotor di.Rk outside its circumference and a flow 0.625 tangent to the rotor dis.k pl8,n . In the case of 11.n 8duator disk, vorticity is shed into the flow at t.he cir umference of the rotor in the form of vortex rings. For this ca~ e 8S "vell as computations with a finite numher of blades, t.he vorticity Rhed from the rotor ips is decomposed into 5 separate vortex tubeR of approximately equal circulation. For

the actuator diRk, the vortex rings entering the flow at each time step are composed of 40 straight segment'. For 1.h discrete rotor bladeR. vortex filaments with one end attached to the blade tipR ar iut,roduced at ea b time, tep. Their length is proportional to the time step itself. The circlllation strength of the vortices is determined b,v co sideration of the total sheacring between the velocity at the inside and 011tRirle of the rotor diRk at its circumference.

The helicopter air frame us d in the study iR representf',d by 7450 trial1<Tles and the ReynoldR DllUlber is 50,000. The )i e volume mesh is 0.005 thick and covered by 11 layers of prisms. The llse of a relativ ly oarse mesh is justifipd by thp g081R of t.his preliminary calculation. In future engineerinr> work it can be anti i ated that consideri'tbly finer surface mesheR that contain more rie ail and aUow for treatment of higher Rpynolds

1ml- IS 'll be 1 il'zed.

7

(a) Sine vipw. (h) Top yi ow.

Figure 2: Vortex filaments produced in quilihrillm by rotor (10 blades) and helicopter body in forwa.rd fli.ght.

Vortex tubes ent. r the calcula.tion both from the actuator disk and from vortir.it.y pro­duced in sa isfaction of the non-slip condit,'on on the surface of the helicopter. The latter tubes enter th flow at the top of the finitp volume mesh covering the helicopter. To a.id in numerical efficiency. vortex filaments that travel ast a plane suffi iently far from the belicopter <'lre removed from the computat.ion.

Particle densities in the cakulatjons are assumed to be sufficienly sparse that the car­rier flow is unaffected by their presencp. In this one-way ('Dupling scenario the pArt.icle motion is d ermined usin.g an empiric:1.l drag law appropriate to turhnlent flow. The only necessary input is a Stokes nnmber, St. Simulations for St = 0 (tracers) to heavy particles with St = 100 have heen computed.

3 Re ult and Disc 10 .

Calculations of the helicopter flow aff~ initiated fr m rest by impulsively adding the constant velocit,ies in the directions parallel and normal to the rotor plime that are

sociated with forward flight. For a typi al compu ation that had 10 rotor blades. equilibrillm conditions in the nnmber of vortex tuhes is reached in approximately 400 time steps covering an elfl.psecl time of 1..5. For the par icular conditions in this case there are approxima.tely 1.5M vorhees I'lt eqllilibrillm. A calculation with the actuator disk reached equilibrium in a simil<'lr tim scale and contained 2.4M vortices. Figure 2 give!'; two view!'; of the vort.ex tubes at equilibrium for t.he 10 blade rotor c<'llculation. The side view Fig. 2(a) , hows the tilted rotor plane from which vortices trail out helicaLl~

from the moving rotor with five vortices appearinO' at. t.h tip of a.ch blade. Contraction of t.he vortex wake is evident 3,.! w II a.."i the eventual degeneration into tmhulence. At the same time vorticit.y formed OVPT the air frame shpds forming a vortical wake. as w 11. A view from above in Fig. 2( shows more clear! the arrangement. of vortices leavinO' t.he rotor blades <'lnd the br AJ(llP of the rotor field strnrture below the helicopt.er body.

8

--02 '-0.3 -0.., -0.5 -0.1~..2 0.1 0

0.5 QA 0..3

0.3

0.2

\

-0"01 01~

0.1

~ a I -0.1

-0.

-04

o~ -0.3 j -0.2

Y

Fi,2,llre 3: Circular planes of tracer particles impactill,2: helicopter jrotor flow field. Parti­cles are uniformly d iRtributed on incoming pl8nes.

Some idea of t.he particlllate field am} how it interR.cts with the helicopter and rotor is iUllstr~Jed in Fig. 3 for a a<; with tm. r parti·J . The effe t of the rotor tIow is to acceler te the parti']es throngh the rotor pl~ne a.bruptly chatl~incr their flow direction. PaTti les impinge on the helicopt.er surface ano are enveloped in t,he smrounding bonnd­arv layers. Consistent with Fig. 2, tmbllient dispernion of the particleR occurs where they int fact with the tmblJ! nt parts of th rotor vort.icitv field a.nd t.he tmblllent wake of the hel icopter.

An int.ereRtincr ftspect. of the particle motions is illustrat. d in Figs. 4 showing the p<'\,ths deriving from sets of particles whose initial positions are on lines within one of the upstream circular disks. The vi W Me, res e tively, from the sine and front. The effe<,t of the motion through the rotor plane in reorienting the partieles is evident in Fig. 4(a). To he ext nt the particle tl'ajec ries show the affects of turbulence it is confined to the reaion most.ly below the helicopter. Thns, particles approach the top part on direct lines while those that interac" with the lower Sll 'faces often timeR appear to be contained within the tnrbnlent flow sllfrOlmclillg the body.

The varil1hility of the mot.ion for particles ent .ring tb _flow at thp s~mp point is illustrated in Fig. 5. HAre, in views from the side and back tbe extent to which the presenre of t.mbnlence disperses t.he inc.oIDing particles is made visible. The iute a.ction with turbnlence means that despit.e the repeatability of the incoming motions, the subsequent motion re ains llnpredictabl. ~nd has th pot.ential to bring parti lp~" to a ran e of possible posit.ions along the RnrfacA of the air frame.

Some indici1.tion of the effe t of Stokes number in how the part.idp..s behave in the virinit\" of the helicopter is iLlllstrated in Ficr . 6. A front, view of t.he particle motions for tracers

9

0.2

0.1

0-'H

-<>.3 -0.1

-0.:1,

-0.2-0.1

~

I -03

" 0.1

-0.4 0.2

-{l.So't I

]-:'=--::"---::'::--.,..,.....--:':-,~'~.,..,.....---::':::---=';:--7'::--:'=---0.6

0.5 0.... 03 0.2 01 0 -Ql -0.1 -0,3 -04 -OS -{l.4 -0.3 -0.2 -0.1 o 0.1 0.2 0.3 0.4 Y z

(1,) Tracers originatin,'" on ,], vertical line. (b) Trrl,cen; originating on flo horizontrtlline.

Fi re 4: Eflect of the flow on tracers.

and for heavy particles wit.h S = 100 are compared for the same flow field. It is seen that there are only relatively minor differences betw, n aths during t.he approach to t.he front. of the helicopter. Subsequent.ly. as t.he particles negotiate around the body the heavier particles ha.ve a cl T endeney t.o traw~l furt.her below the surface before being affected bv the turbulence. The effect of the latter is lITeater on the tracer particles. In fact. a side vie\" would show that their t.rajectories are more likely to trail along para.llel to the lower. 1.lfface.

To6nd ant the effect of sensor placement on t.he efficien y with whieh the llpstrp.A.ID atmosphere can be sampled for particles it is nee Rary to ain information ahout the initial location of particles that arrive itt given points on the helicopter. For each of the nodal point.s that make np the helicopter surfa~e, th colle tion of partides arriving within a small distance (0.01 in the present case) were computed. For ea.ch of these pat.hs, the initial point where th partie! came from within the circular disks of particle points was c1eterminecl. From tilic:; the maJUmllm distance between the initial points wa~g

computed. Fi J.re 7 shows the percent-a e of location~ on the airframe ~nrface for which arriving part.icle~ have sta.rted ant at va.rying distances. For approxima.tely 65% of the surface points partic1e.R arrive from jnst a single npst.ream loca.t.ion. For a. much sma.ller percen O"e of t.he smface points the part.icles arrive from positive distances apa.rt. A very RrnaU percenta e of locations are distin!ruis ed hy th fact that the particles ~.rriving there hi1,ve begun from relatively far distances apart, in fact., on the order of more than half a rot r diame

10

-Q.L

0.4

0.3

0.2

0.1

01-------+-----­

H -0.1 -<).3

-0.2

-0.3

-0.4

OA -0.5

os

-0.2 -0.1 o 0.1 0.2 0.3O.6L--:-o~5-----'-<JC-:_.----:;-o;":;3----:;-0;":;'----:;-0:::_'--;;-0-:':0.1'---:':02;--~03;---;;O_':---;;O-;-5­y z

(H.) Sirl.e view. (h) Back view

Figure 5: Paltid paths with common ori,gin.

1

0.4

0.3

0.2

0.1

Ol-------I~-----

H -0.1

-0.2

-0.3

-0.4

-0.5

0.4

0.3

0.2

0.1

Ol--------lr------­

H -0.1

-0.2

-0.3

-0.4

-0.5

-0.2 -0.1 o 0.1 0.2 0.3 -0.2 -0.1 o 0.1 0.2 0.3 z z

(H,) St = O. (b) SI = 100

Figure 6: Effect of Stokes number on pa.rticle tm,iectories.

70.---~-----r--,----~-~_-~---,

60

'0

Fignre 7: Percent<,\o-es of the airfram snrface area for which arriving; particles ba.ve different rnagmtllde. of initial se .aration.

12

-0.Q.5

-035

-0.4 '----.-;;';0';---:.07' --;:----,0"';-'-7.0.2;---;O~.3--;;'O.<;---;O~.5--:!06

Fi,gme 8: Surface locations for which particles I\uive from disparate initial points: red, dist(lJ}(' s > 0.25: hille, rlist.Anc s bet.ween 0.17.5 and 0.25: !IT ,no distances h ween 0.1 ann 0.175.

The plot in Fig. 8 indicat.es the physic-allocations on t.he helicopt.er where particles arrive from t.h larg t initi I separatiouf'. Particles h ginning more than 0.25 apArt, denoted in red. are clustered at a few points, SllITOlindin th m in bille are particles that have stArte out be ween 0.17.5 and 0.2.5 apart. There is little doubt from the result. h wn here that the tail section of th helicopter provides a more optirnallocation to efficiently sample upstream particles than the main part of the helicopter.

An explanation for the result in Fig. 8 is given in Fig. 9 showing the collection of part,icle trajectories t at arrive at t e nrface points incli . ted i r d in t e former figme, (i.e. the initia.l particle location are as much as 0.25 apart). The particle paths in question e s en to arrive at th tail section from two different mechanjsms. In the first. hey travel directly throll h the rotor plane t.o the t.a.il region. In the second, theyflrst approa'h the main s dion of the heHcopt.er and th n get swept np in the turhllient wake flow that carri s them towarrl tb tail s· -ti n. T i.' b i meeh is WfL'3 observed in a number of different simulations. for example, compntations l1sing the act.uator disk model and for ma$sed particles with a range of St kes numbers. Further consideration of how to optimiz the placement of a particle sensor is appropria.te once a specific helicopt.er design is to be considered. In t.hat, case. using a high resolut.ion t.riangularization of t.he snrface. showing important local featllT ~'i. the pr sent code ma.y be execllted to gain comprehensive information about t.he PFlItic! paths. In such work it would be appropria.te to investi ate the e}...'pected performanc of a collector in a ranO'e of flow onditions.

4 Co clus·o s,

A positive answer has been rea bed to the prima.ry 'Inestion posed in this St.11dy to the effect that the positioning of a particle collector on unmanned h licopters can potentially have a signjficant effect on the efficiency with which the atmosphere can be scanned for hazardous piles. For a numher of test cases consider here the same basic

13

, 0"'

I -0

-o.S

-0.3

-().2

-01

., 0.2

0.3

OA~----r---'--~~~~ 05 0.8 08 0.4 11.2 ~a..2 -0.. -0,6 -09 -1

Y

Figme 9: Paths arriving at commOn locat.ions on t.he helicopt.er sllTfa,~(' for which the init.ial separa.tion is > 0.25.

physical phenomena W<1R observed: part.icles t.hFlt arrive at the tt!jl section of the helicopter airframe have a Teater likelihood of having come fr ill disparat.e upst.ream positions. In effect, turbulence shed 0-[ of the helicopter in fonvR..rd flight brings. particles to locations in the ba.ck itt t.he tail where they join particles that have directly itppr ach d the titil region thrau h the rotor field. For the C&qe of tracers and a rot.or with 10 bltt.des t.he 11 s n s paration of the particles w s grea.t.er than balf t,he rotor radius. Statistics of the same order were fonnd in oth r Ct!. e, t.reat.ed. Detailed predict.ions produced by high resolution simulations will be ppropriate when particular air frame sha.pes, rotor confignration, and flight conditions are investigat d.

The simul... tions a complish d in this Rtudy are made possible by the nniqlle qnalities of the numerica.l scheme that Wtt.'i a )plied and developed in this project. The minimFl.I gridcling requirements in which . 1st a surface trian ularizat.ion is necessary allowed for rapid progress in implement.ing t.he methodology. The LagrRn<Tian nat.ure of the filament

cula ion dovetailq w 11 with t.he La Tangian particle computation,. oreover, the capR.hility for capt,llring t.he dynamic:::: of sb d v rt.ical ::::truct.llTe provides a good ba.'iis for simulating tbe partide dispersion that is nece, sary for reaching conclusions in the project. In contrRflt to the p pnlflrly formnlated free vort.e.,,< models of the rotor field that Q'enerally reI r on the use of a single vortex streamin from th rotor tips [17]. the present. fiimnlation lltiliz s an arbitrary Dllmber of filam nts for each rotor - five, in the present eomplltations. The us of mt Itiple vortiees 8110ws for the r presentation of vortical strllctme that is ess ntial to the physical phenomeuA. being moneled. Moreover, solving for the rotor tip vortices aB par of the g nera.l vortex filament algorithm allows for consideration of practi al scenarios in which the inflnence of the helicopter bo . r

flmong many other fact rs is present. The conchlsion rna. be drawn that the teebnolog,\' developed in this project hIlS the pot ltial for much wider applica.tion than the spe ific problem considered here. In particular, it is not hard to envisage that the VorCat 8eb me

14

5

can offer new capabilities in the design and analysis of heli opter flows.

Recomme d Hons.

Om objectives for a Pha..<;e II continuation of tbis project R,Te:

• Ose VorCR,f, simulations to optimize l'lno eVl'l,lnate the performance of pl'lTticle col­lection technologies developed at Boeing for unmanned platforms including the Hummingbird rotorcraft and ScanEagle n.xed-wing.

• Continue the develop ent, validation and application of the VorCat platform to a range of related problems of interest to government and industry for which chem/bio hazards may be present.

The p sitive results of this work have raised the interest of our partners at Boeing in applying the present algorithm to studying flows associated with Boeino--specific DAV designs. In these cases, the use of high resoilltion surface grids can be employed that will allow for many of the details of the flow field to be accounted for nmnerically. If also a particular tvp of airborne particle is considered then, imula ions around realistic air frames usino- reaJistic particles can be Ftccomplished. Addi.tionally, a mnge of possible flight conditions an be considered. This kind of engineering study can be coupled with physical experiments (wind tunnel and/or flight tests) carried ont by Boeing during Phase II and/or Phase HI work yielding a means for optimizing the particle collection technologies employed on Boeino- aircraft.

The continuation of our Phase I study into Phase II will allow VorCat, Inc. to (i) complete t.hf~ development and testing of a unique, valnable flow simulation tool that will have a high probability of being purcha.<; d by a major DOD contractor such as Boeing, (ii) prove the superiority of the VorCat software over other currentTy avai1abte technoloe:ies for use in similar applications and. consequently, (iii) provide VorCat with the opportunity to extend its reach into other DOD and industrial applications that are closeIv related to the applications rliscllssed in this report.

References [lJ Anderson, G. P. et al. "Biological agent detection with the use of an airborne biosen­

sor," Field Anohlt. Chern. Tf'chnol. 3, 307-314, 1999. [2] Bernard, P.S. and Wallace, J.M. Turbulent Flow: Analysis, Measurement and Pre­

diction, John 'Wiley & Sons. 2002.

15

[3J Bernard, P. S., Collins: .J. P. and Potts, 1\1L "Vortex method simulation of ground vehicle aerodynamics," SAE Transactions Journal of Passenger Cars - Mechani al Systems, 612 - 624: 2005.

[4) Bernard, P.S. "Thrbulent flow properties of large seal vortex systems,''' Proc. Nat. Acad. Sci., 103, 10174-10179, 2006.

[51 Bernard. P. S. "Gridfree simulation of the spatiallyo-fOwing turbulent mL'Cing layer," AIAA .1.,46, 1725-1737: 2008.

[6J Bernard, P. S. "Vortex filament simulation of t.he turbulent coflowing jet," Phlls. FhLid.s, 21:025107, 2009.

[7] B rnard. P. S.: Collins. P. and Potts. M. "Vortex filament simulation of the turbulent boundary layer." AIAA-2009-3547. presented at th 19th AIAA Comput.ational Fluid Dynamics Con ress. San Antonio. Texas. June, 2009.

[8J Bernard, P. S., Collins, P. and Potts. M. "Vortex filament simulation of the turbulent boundary layer," AIAA 1., to appear, 2010.

[9J Chorin, A. J., "Hairpin removal in vortex interactions II," 1. Comput. Phlls. 107. 1 - 9. 1993

[10] Chorin, A. J., Vorticity and Turbulence: Springer-Verlag, 1994. [11] Fessler, J. R., Kulick, J. D. and Eaton, J. K. "Preferential concentration of heavy

particles in a turbulent channel flow," Phys. Flu'ids 6, 3742 - 3749, 1994. [12] Goodin lT, J.J. "Biosensor technology for detecting biological warfare agents: Recent

progress and future trenos," Analytica Chimica Acta. 559, 137-151,2006. [13] Green ard, L. and Rokhlin, V., "A fa.'lt algorithm for par ide simulations." .I. Compo

Phys. 73. 325 - 348. 1987. [14] GillSU!. 1., "Vortex flows on AVs: Issues and chall n..,;es." Aeronaut?:cal J. 108, 597

- 610, 2004 [15] Hill. A., "Modeling concentration profileR of chemical warfare agents to assess the

usefnlness of airborn chemical det ctors," Defense Science and Technology Organisa­tion. Technical Note DSTO-TN-O. 06. 2003.

[16] HOlTan, R. C. and Cuzzi, J. N. "Stokes and Reynolds number dependence of prefer­ential pa.rt.icle concentration in simulated three-dimensional turbulence." Phys. Fluids 13. 2938 - 2945: 2001.

[17] Leishman. J. G., Bhagwat., rvI. J. and Bagai. A.. "Free-vortex filament methods for the analysis of helicopt I' rotor wakes," .I. Aircraft. 39; 759-775. 2002.

[18] Lynn, R. and GursuL 1., "Vor e dynamics on a generic UCAV COnfi,gllIation." ATAA-2006-61 44th AIAA Aerospac Sciences Meet.ioo· and Exhihit. Reno, Nevada. .Tan. 9-12.2006.

[19] Naimllshina, A. N. et a1. "Airborne analyte detection with an aircraft-adapt.ed sur­face plasmon resonance sensor system," Sensors and A /;uators B: Chf'miml104 237­248,

[20] air. M. L. "Cone ptllal desi,g}). engineering modelil IT. and experimental validation of air sampling syst.em for chemic:al sensor insert.ion into the U.S. Army Research

16

Laboratory's (ARL) silent operating aerial reconnaissance (SOAR) program." ARL Report A963124., 2004.

[21J Puckett. E. G.. "Vortex methods: an introduction and survey of selected research topics:' III M.D. Gunzburger and R.A. Nicolaide , editors, Incompressible comp1J,ta­tiona.l jiu'id dyna.m.ics: t1' nds and advan es, Cambridge Cniversity Press, 335 - 407, ]f)Q'3.

[22] Squires, K.D. and Eaton, J. K., "Preferential concentration of particles by turbu­lence," Ph7js. Fluids A 3, ] 169. 1991

17