SAD-A269 698

WL-TR-93-3047

COMPUTATIONAL FLUID DYNAMICS

(CFD) RESEARCH BRANCH

TECHNICAL BRIEFS

Don W. Kinsey et al.

SEP2819,

15 May 1993

Final Report for Period May 1992 - May 1993

Approved for public release: distribution is unlimited.

93-22339FLIGHT DYNAMICS DIRECTORATE

9i 3 - 2339I ,, , ,

WRIGHT LABORATORY I I' i ubuAIR FORCE MATERIEL COMMANDWRIGHT-PATTERSON AIR FORCE BASE. OHIO 4.5433-7,562

U .-- ' ? • -

NOTICE

When Government drawings, specifications, or other data are used forany purpose other than in connection with a definitely Government-relacedprocurement, the United States Government incurs no responsibility or anyobligation whatsoever. The fact that the government may have formulated orin any way supplied the said drawings, specifications, or other data, is notto be regarded by implication, or otherwise in any manner construed, aslicensing the holder, or any other person or corporation; or as conve'i ngany rights or permission to manufacture, use, or sell any patented inventionthat may in any way be related thereto.

This report is releasable to the National Technical Information Service(NTIS). At NTIS, it will be available to the general public, includingforeign nations.

This technical report has been reviewed and is approved for publica-

DON . KNSE, Teh M JOEPHM. MANTERInterdisciplinary and A plied CFD ChiefSection CFD Research Branch

1 L." 'Jd1LaDOHIIS SZDLOCKActing ChiefAeromechanics Division

If your address has changed, if you wish to be removed from our mailinglist, or if the addressee is no longer employed by your organization pleasenotify WLIFIMC , WPAFB, OH 45433-2U22 to help us maintain a currentmailing list.

Copies of this report should not be returned unless return is required bysecurity considerations, contractual obligations, or notice on a specificdocument.

DDISCLAIMER NOTICE

THIS DOCUMENT IS BEST

QUALITY AVAILABLE. THE COPY

FURNISHED TO DTIC CONTAINEDA SIGNIFICANT NUMBER OF

COLOR PAGES WHICH DO NOT

REPRODUCE LEGIBLY ON BLACK

AND WHITE MICROFICHE.

Form ApprovedREPORT DOCUMENTATION PAGE oj8N 07o041

~JT e .3 r... A, a "e j G~nrOo Ir'd .. roU.iVc . . * 1-1. * -)-at .3 C d : ntn fWI4'0.Pg T?`- b~fdtf fttmate of .nv Athet 4%00< of ftfl),i 0IIIR w,1 I I lte cHeo.c ~..o,'" • •:]•"•' ! " .a, s:,le,- I' -, * ..•rr,- 3 +• ,ma,., . *r•,4 'K -V ,sr ,,'* •o ',•o • •,' •, .•r ' -e • * o' '•+ ",",, " Oor, I-,on'• de'd,• ~E.g+n l 'u ta~ r

'.,.~~~~ jo * 1 iw uf 4 0u T,r t~'. I'* pl-. t~o 0,'d lede ~ sO?4O ) b~'qsr' ( 2A050

1. AGENCY USE ONLY 'Leave blank) 2 REPORT DATE 3. REPORT TYPE AND DATES COVERED

19 Jun 93 Final Report May 92 - May 934. TITLE AND SUBTITLE 5. FUNDING NUMBERS

Computational Fluid Dynamics Research Branch Technical PE: 62201F

Briefs PR: 2404TA: 10

6. AUTHOR(S) WU.: Al

Don W. Kinsey, et al.

7. PERFORMING ORGANIZATION NAME(S) AND AODRESS(ES) 8. PERFORMING ORGANIZATION

Flight Dynamics Directorate REPORT NUMBER

Wright LaboratoryAir Force Materiel Command WL-TR-93->'-+7W+;ight-Patterson AFB OH 45433-7562

9. SPONSORING/ MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSORING/ MONITORING

Flight Dynamics Directorate AGENCY REPORT NUMBER

Wright LaboratoryAir Force Materiel Command WL-TR-93-3047

Wright-Patterson AFB OH 45433-7562

11. SUPPLEMENTARY NOTES

Technical Summaries of in-house work completed during the previous year. Workcovers both basic research (6.1) and research and development (6.2).

"12a. DISTRIBUTION /AVAILABILITY STATEMENT 12b. DISTRIBUTION CODE

Approved for public release; distribution unlimited.

13. ABSTRACT (Maximum200 words)

This report consists of brief technical descriptions and one or more pictures of

charts depicting research conducted in-house during the past year. The briefs

are designed to provide enough information to clearly define the task, but are

short enough to hold the reader's interest. Where applicable, a reference for

more detailed information is provided. The research covers a wide spectrum ofwork; from grid generation to flow solvers, to post-processors; and from

incompressible to hypersonic speeds.

0:j~j c -m.&zia colorFCi-A)41 Aai DYIC reproduct-Vý-F• nill be 1t black aw

14. SUBJECT TERMS 15. NUMBER OF PAGES

Computational Fluid Dynamics, Grid Generation, Hypersonic Flow, 31Unstructured Meshes. 16. PRICE CODE

17. SECURITY CLASSIFICATION 18. SECURITY CLASSIFICATION 19. SECURITY CLASSIFICATION 20. LIMITATION OF ABSTRACT

OF REPORT OF THIS PAGE OF ABSTRACT

Unclassified Unclassified Unclassified ULNSN 7540-01-280-5500 Star-dard form 198 (Rev 2-891

S, r,-o ,' .a,.:f f

GENERAL INSTRUCTIONS FOR COMPLETING SF 298

The Report Documentation Page (RDP) is used in announcing and cataloging reports. It is importantthat this information be consistent with the rest of the report, particularly the cover and title page.Instructions for filling in each block of the form follow It is important to stay within the lines to meetoptical scanning requirements.

Block 1 Agency Use Only (Leave blank) Block 12a. Distribution/Availability Statement.Denotes public availability or limitations. Cite any

Block 2. Report Date Full publication date availability to the public. Enter additionalincludingday, month, and year, .favailable(e.g limitations or special markings in all capitals (eg.Jan 88). Must cite at least the year NOFORN, REL, ITAR)

Block 3. Type of Report and Dates Covered. 0O See DoDD 5230.24, *DistributionState whether report is interim , final, etc. If S e e o n Technicalapplicable, enter inclusive report dates (e.g. 10 Statements on TechnicalJun 87 - 30 Jun 88), DOE See authorities.

Block 4. Title and Subtitle. A title is taken from NASA - See Handbook NHB 2200.2.the part of the report that provides the most NTIS - Leave blank.meaningful and complete information. When areport is prepared in more than one volume, Block 12b. Distribution Code.repeat the primary title, add volume number, andinclude subtitle for the specific volume. Onclassified documents enter the title classification DOE - Eave blank.inparentheses. DOE -Enter DOE distribution categoriesin pfrom the Standard Distribution forBlock 5. Fundinq Numbers. To include contract Unclassified Scientific and Technicaland grant numbers; may include program Reports.element number(s), project number(s), task NASA - Leave blank.number(s), and work unit number(s). Use the NTIS - Leave blank.following labels:

C - Contract PR - Project Block 13. Abstract. Include a brief (MaximumG - Grant TA - Task 200 words) factual summary of the mostPE - Program WU - Work Unit significant information contained in the report.

Element Accession No.

Block 6. Author(s). Name(s) of person(s) Block 14. Subject Terms. Keywords or phrasesresponsible for writing the report, performing identifying major subjects in the report.the research, or credited with the content of thereport. If editor or compiler, this should followthe name(s). Block 15. Number of Pages. Enter the total

number of pages.Block 7. Performing Organization Name(s) andAddress(es). Self-explanatory. Block 16. Price Code. Enter appropriate price

Block 8. Performing Organization Report code (NTIS only).Number. Enter the unique alphanumeric reportnumber(s) assigned by the organizationperforming the report. Blocks 17. - 19. Security Classifications. Self-

explanatory. Enter U.S. Security Classification inBlock 9. Sponsori nq/Monitoring Aqency Name(s) accordance with U S Security Regulations (i e,and Address(es) Self-explanatory UNCLASSIFIED) If form contains classified

information, stamp classification on the top andBlock 10. Sponsorinc/Monktorincg.A encjg__c bottom of the pageReport Number (If known)

Block 11. Supplementary Notes Enter Block 20. Limitation of Abstract This block mustinformation not included elsewhere such as be completed to assign a limitation to thePrepared in cooperation with . Trans of , To be abstract. Enter either UL (unlimited) orSAR (samepublished in. When a report is revised, include as report). An entry in this block is necessary ifa statement whether the new report supersedes the abstract is to be limited. If blank, the abstractor supplements the older report is assumed to be unlimited.

Standard Form 298 Back (Rev 2-89)

Contents

Introduction ........ .......................................

Three-Dimensional Unstructured Grid Generation .................... 2

Naiver-Stokes Simulation on Unstructured Hexahedral Meshes ......... .3

COBALT - An Unstructured Flow Solver ..................... 4

Unstructured Approach to the Design of Multiple-Element Airfoils 5.

Three-Dimensional Unstructured Post-Processing .................... 6

TOPDUUG - The Only Package for Design Using Unstructured Grids . , 7

Numerical Simulation of Turbulent Cylinder Juncture Flowflelds ........ 8

Turbulence Modeling for Thrust Reverser Flow Prediction Methods . .. 9

The Effect of Leading-Edge Cross-Sectional Geometry on Vortex Flow Aero-

dynam ics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10

Vortex Breakdown Above a Pitching Delta Wing ................. 11

Numerical Simulation of Delta-Wing Roll ..................... 12

Nonequilibrium Hypersonic Flowfield Prediction for Flow Past Blunt Bod-

ie s . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

Stability Analysis of a Combined Couette-Poiseulle, Two-Fluid Flow . . I I

Computational Aerodynamic Analysis of a Decoy Configuration ....... 15

Euler Analysis for Delta Wing Design .... .................. 16

Numerical Simulation of a Jet Expulsion from an Aircraft Forebody, With

Comparison to Experiment . .............................. 17

iii

Numerical Analysis of a Chined Forebody with Asymmetric Strakes . . .

Iuviscid Hypersonic Flow Around a Conically-Derived Waverider ..... .. 1

The Flowfield Past the X24C Reentry Vehicle .................. 20

Calculations Over a Swept Wing with a Gap in High Enthalpy Air .... 21

Bo mnb Blast Sim ulation ................................ .

Parallel Processing for Computational Fluid Dynamics ............ 23

Advanced Technology Air Vehicle Geometry Database ............ 24

Computational Electromagnetics .......................... 2.5

List of A uthors .. .................................... 26

Aacesslion PorNTTS iI&DTIC T;-F 11 00ElC

EYI•C UAL•,JISPB ;:•lF'

ix

IntroductionDon W. KinseyCFD Research Branch

This technical report was produced by the Computa- The technical descriptions provided in this report fol-tional Fluid Dynamics Branch of the Aeromechanics low closely the format used in NASA's National Aero-Di'ision. Flight Dynamics Directorate, Wright Labora- dynamic Simulation annual report. The articles are de-tory. The CFD Branch is composed of two Sections; the signed to provide enough information to clearly defineCFD Research Section. and the Interdisciplinary and the objective, approach and significant accomplishmentsApplied CFD Section- The CFD Research Section con- of the task, but be short enough to hold the reader's in-ducts basic research which is largely sponsored by the terest. Where applicable, a reference that contains moreAir Force Office of Scientific Research. The IACFD Sec- detailed information is provided. In all cases, the nametion conducts research and development of a more ap- of the author is provided and he may be contacted di-plied nature. Thus the CFD Branch as a whole covers rectiv for details about his work. An alphabetic listinga very wide range of technologies from geometry gener- of each author, his phone number, mailing address andation to post-processing of CFD results. e-mail address is provided on the last of the report.CFD technology is expanding very rapidly. New algo- The picture chosen to go with this introduction is rep-rithms. techniques and applications are being added al- resentative of our current capability in applied CFD.most daily. Some of these represent significant advances The geometry was digitized from blueprints, the dif-in the state-of-the-art, while others are of value to a ferent parts of the geometry were combined, and themore limited audience. Most of the more complete re- surface grid and block faces were produced using oursuits are published in symposium papers and journal I3G/IVIRGO interactive graphics package. The volumearticles, but many smaller efforts are never published. (3-D) grid was computed with a code called PLUTO. AnIn either case it is easy for the work to htcome lost: the Euler solver called MERCURY. based on Jameson's fi-published work is lost in the volumes of papers and ar- nite volume technique, produced the flow solution. Thetides that we never seem to have time to read. and the surface geometry and particle traces were produced withunpublished works are lost in the files of the researcher. the NASA-developed PLOT3D graphics package.The objective of this report is to document in a singlesource the preceding year's CFD accemplishments.

Euler Solution on a Complex Fighter G,Eomeiry

• • I I1

Three-Dimensional Unstructured Grid GenerationFrank C. Witzeman JrInterdisciplinary and Applied CFD Section

Research Objective body) and the enforcement of triangular faces which lieOne of the most difficult and time-consuming tasks in on the geometric surfaces. Further advantages of thethe CFD solution process is grid generation. Interac- method include the automatic insertion of grid pointstive graphics-based programs such as GRIDGEN and in regions of interest 1adaptive gridding) and relocationI3G ,/VIRGO have overcome many of the limitations and of grid points for moving-body simulationslabor-intensive tasks of multiple-block structured gridgeneration. Although these methods (and many others) Accomplishmentshave enabled CFD solution turnaround times on the or- A preliminary version of the unstructured grid genera-der of weeks or days. they cannot meet design environ- tor. TETMESH. has been produced. The software is be-ment goals of days or hours. This effort is focused on re- ing evaluated by several Air Force organizations. Three-ducing CFD solution times from weeks to days through dimensional unstructured grids have been constructedthe development of a highly-automated. easy-to-use un- for several test cases. and Euler solutions have been ob-structured grid generation procedure. This method will tamed using our in-house flow solver COBALT. For thebe applicable to arbitrary and complex vehicles and example illustrated below, approximately 1 week wascomponents which are nearly impossible to model with required to generate a multiple-block structured grid.the structured grid methods. For the same configuration- only 6 hours were needed toApproach generate the unstructured grid.

The unstructured grid generation method is part of the Future Plansoverall Multi-Bodied CFD Method being developed un- The interactive three-dimensional unstructured gridder contract to the Computational Mechanics Company generator will be modified to be more robust and user-(Dr S. R. Kennon, Principal Investigator). The method friendly. More generalized criteria will also be developedis based on a generalized three-dimensional Delaunav for unstructured grids made up of irregular tetrahedralconcept which produces grids comprised of tetrahedral and triangular-prismatic elements. These types of gridselements instead of structured, hexahedral elements. are required for accurate simulation of viscous flows.For most cases. the only input required is an accurate Publicationsdescription of the geometry. Currently. this geometry Kennon. S. R.. et al. Geometry Based Delaunay Tetra-description is the same as that used for the structured- hedralization and Mesh Movement Strategies for Multi-grid methods. Unique features of this method are the Body OFD." AIAA Paper 92-4575. August 1992automatic addition of grid points (both on- and off-

Vnstructured 4rid and Pressure Contours on a Decoy Body and Planar Slice

Navier-Stokes Simulation on Unstructured Hexahedral MeshesCaptain Michael J. AftosmisCFD Research Section

Research Objective dated on a variety of inviscid and viscous test cases. Nu-This research effort develops and evaluates an adaptive. merical simulations of 3D supersonic corner flows, sub-high-resolution, upwind algorithm for the simulation of and transonic airfoil flows and simulation of laminariiviscid and viscid flows. Upwind methods provide dis- flow over a delta wing were all in agreement with e.xper-crete representations of shocks and contact discontinu- imental and theoretical results The method typicallyities using as few cells as possible. and adaptive mesh requires 10 to 20 times fewer cells than all equivalentlytechniques automatically adjust the local flow physics. resolved solution on a structured mesh. Moreover. si51ceThe combination of these two approaches suggests a the number of operations which must be performed atnatural and extremely flexible platform for simulating each time step is proportional to the number of o46l.-flowfields with features spanning a variety of disparate each iteration requires less computer time. Thus. over-length scales. This work seeks first to overcome some of all turnaround time is greatly reduced.the practical and implementational issues involved with Future Plansthe development of such a technique. and then to evalu- The method will be applied to time-dependent prblenmsate the utility of the resulting method in its application including those containing significant geometric cointo a variety of prQblems of interest. piexity. Evaluation of the performance of the techniqu,:Approach for such problems is essential before the technique willThis effort applies a Roe-based. finite-volume, upwind be considered mature.technique on an adaptive mesh consisting of unstruc- Publicationstured hexahedral cells. Viscous terms are modeled with Aftosmis. M.1 . "An Upwind Method for Simulation ofcompact central differences. -As the solution evolves, the V.iscous Flow on Adaptively Refined Meshes." Acepted.tniesh cells adapt through directional cell division in or- for publication by the A.AA .ournal. Mar. l19V3.der to increase the resolution of the computation in and Aftosinis. M.J.. "Viscous Flow Simulation I.-:imgaround emerging flow features. By concentrating new and Upwind Method for Hexahedral Based Adaptivecomputational nodes only where the solution requires Meshes." AIAA Paper 93-0772. Jan, 1993further enhancement. the method uses far fewer cells Aftosmisk. ..J. "A Second-Order 'IVD Methvd forthan comparable structured mesh techniques. the Solution of the 3D Euler and Navier-Stokes Equa-

tions on Adaptively Refined Meshes." Proccadings of thiAccomplishments iI3th Inlernational Conference on Nrumerical IIt~hd,, mAn unstructured computer code was written and vali- Fluid Mechanics. Rome. Italy. July 1992.

Pressure contours for viscous supersonic flow over cropped delta wing. T7e7 final adapted mesh contains 999.000 nodes.The inset shows a comparison with wind tunnel data.

COBALT - An Unstructured Flow Solver\Villam Z $trangInterdisciplinary and Applied C'-D Section

Research ObjectiveTo simplify computaional fluid dynamics (CFD) anal- Second- and third- order accuracy in time is achievedyses of air vehiciet- uf arbitrary geometrical complexity: with classical Runge-Kutta integration schemes Mono-to lessen the ,ime required to generate such predictions tonicity, required for stability of the spatially second-by a fact,ýc of -: to reduce the CFD skill level required order scheme. is based on the works of Collela. Goor-of th•i ngineers conducting the analyses. jian and Obayashi. Slip wall. adiabatic and isothermalA !,,.,,roach no-slip wall. farfield, inflow and outflow boundary condi-"A octions are -. ailable. Considerable effort has been devotedUnstructured grids are one of the very few methods, if to these boundary conditions to make the code as robustnot the only method. offering the potential to achieve and as accurate as possible. Construction of the viscousthe above research objectives. In theory. a user gener- terms follows MacCormack's work. The Baldwin andates only the bounding surfaces of the problem and the Barth turbulence model is used to simulate the effects ofunstructured mesh generator automatically tessellates fine scale turbulence. Two-dimensional. axi-svynmetricthe domain interior. LV nstructured grids, however. re- and three-dimensional problems may be treated. Thequire flow solvers, that are capable of handling general sole requirement of the mesh is that it be composed ofconnectivity issues. Thus. an in-house project to de- convex cells. Lastly. with the exception of a few pre-velop an unstructured flow solver, COBALT. began in and post-processing tasks. COBALT is fully vectcrizedFebruary 1P00, and typically operates in excess of 100 MFLOPS a Cray

Accomplishments XMP,2 lThe fundamental algorithm of COBALT is the spatially Future Plansand temporally first-order accurate, non-linear schem- Additional validation of the viscous and turbulencedue to Godunov The extension t~o second-order accu- model routines, documentation. third-order spatial ac-racy in spa,-,u is d eue o to van Leer and Collela. curacy, parallel implementation. additional turbulencemodels and grid adaptation are planned.

"Iiurbulpnrt lditimn of an Axi-Syrnumnrri, Plut.' Nozl,z

Unstructured Approach to the Design of Multtiple- Element AirfoilsDon W. KinseyInterdisciplinary and Applied ('ED Sectlon

Research ObjectiveDevelopmient of efficient high-lift airfoils may lead to Accomplishmentsredluced weighit and comnplexity over current high lift, The programn was developed and validated on severalsystemis witii attendant reduced cost and improved re- single-element airfoils. The first multiple-elemient airfoilliability. The research described here combines the rel- test case was a WNhitcomb slotted supercritical airfoi!ativelv new unstructured grid computational fluid dy- with a trailing-edge flap set at 2 0 '- The target velo -

namics (CFD) technique with an existing inverse desi' itv distribution was obtained [,v direct Euler solutio-n jfProceduire. The results will. hopefully. be a new tech- th~e known 'Whitcomb geometry. The Initial airfoil -nique to diesign and analyze high-lift ai'rfoils with the inm- omletry was two symmetric (NACA 0014) airfoils withproved accuracy of an Euler solver. but significantly less chr eghadtaln-de locations of the target g-pwork for the deýsigner than most advanced CFD meth- ometries. The picture below shows the initial and finalods- geometries and pressure distributions, .A three-elenwnt

Approach airfoil test case which had a :30' leading-edge flapý and a.30' traihng-edge flap was attempted. but proved -,c be

The ('ED solver ýVas an implicit Van Leer split flux Eu- too difficult a case for the current method.ler scheme with point Gauss-Seidel relaxation for two-dimensional triangular g'rids- A two-dimensional un- Future Platisstructuredi grid generator bascd on the Delaunav trr- A three-dimentional version of the Inverse Jesign proce.angulation process was used to create the grid. Grid dure will be developed. The same MGNM metho-d andmovement. needed to allow the geometry t~o change with srn nlg ilb sd oee turrdhi~the inverse desion procedure. was provided by a '-sprin- volume Euler solver based on Jaineson s alzoritha ilanalogy" process The inverse design program is a vari- be used for the flow solver. This pro,-ram will r.-.ui--ation of the 'Modified Garabedian-McFadden (NIGN- a two-block structured grid that is allpined strearinwi--residuai-correctioii procedure-. This procedure requires ovrtewn.Aimlwn'bdtstcewllea rareot veloritv distribution and an initial airfoil g-used t~o develop and validate the codeonietry. Then. through a series of iterations, the airfoil Pbiaingeomemtry Is adjustfed based on the differe-nce between -inv DAV a ;nd .Jolly. B.A..An Vi nstructur'~i Ap-the curreýnt and taroet velocity distribution., The con- prahtth DeinoMutleEmntAro.vergied solution proxides a geornetry that produces the AIAA Paper 92-2709. June 1992.t.irg~et velco-it.% distribution.

INITIAL CONDITIONS FINAL CONDITIONS

TARGET SOLUTIONIIILSOLUTIONC

0p -- 1NT

Three-Dimensional Unstructured Post-ProcessingLt Phillip %M. FiteInterdisciplinary and Applied CFD Section

Research Objective larger unstructured grid size This Increase.f, availableFollowing grid-generation and flow-solution, the third memory by reducing the need to calculate connectI'-critical stage to the CFD solution process involves ity data except when needed Incorporating memorypost-processing. or the visualization of calculated data sharing techniques such as computing derived quantitiesPost-processing provides scientists and engineers with a (Mach number.entropy. etc *on the fly' eliminates thegraphics-based analysis tool for presenting CFD data a need to carry those values in the processing arrayswell as verifying the grid-generation and flow-solver ap-plications. Due to their simpler generation and quicker Accomplishmentsadaptation characteristics, unstructured grids are be- With the current options available. VPLOTI3Ds un-coming the choice in the CFD community as the need structured capability has improved memory allocationto analyze more complex geometries increases. To- to process grid sizes of up to 2.50.000 grnd points Inday. there are many post-processors for structured grids addition to grid size input, working versions of the rodeavailable, yet few meet the current demands for visual- possess some unique features for unstructured process-izing unstructured grid data. ing. These include arbitrary cutting planes iso-volumeApproach plotting, and input of multiple data formats.

An unstructured post-processing capability is being de- Future Plansveloped under contract to Vigyan, Incorporated (Dr. To meet future Air Force needs. V'PLOT3D will be mod-Paresh Parikh. Principal Investigator). Originally devel- ified to incorporate some additional features The-e in-oped for NASA. Vigyan's code VPLOT3D was the only elude portability to multiple platforms. tinme-dependentavailable post-processor capable of visualizing unstruc- visual simulation ianimation) of unstead. flows ardtured 3-D grids. Original versions of this code were lim- moving objects. plotting time dependent data on a mo'-ited to processing unstructured grid sizes up to 180.000 ing plane or grid. plotting experimental data. and storegrid points. To allow the input of larger grids. Vigvan separation analysisis modifying VPLOT3D by increasing the code's mem- Publicationsory management capability and incorporating memory Parikh. P. et al. "A Package For 3-D Unstructured Gridsharing techniques. To improve memory management. Generation. Finite Element Flow Solution and Flowstructured grids are handled separately from unstruc- Field Visualization," NASA Contractor Report 1.8209%.tured grids. Separating grid types reduces unnecessary September 199.data associated with structured meshes, a-i thus allows

High Speed Civil Irin.,port with ('p mnt:n.irs in a 3-D [.nstru'tutredl Grid

6

TOPDUUG - The Only Package for Design Using Unstructured GridsMarvin C, GridleyAirframe Propulsion Weapons Integration Section

Research Objective 3. Flow solver job file generationThe use of CFD in research and design can often b? a 4. Flow solution visualiza : 'ntime-consuming and somewhat esoteric "arr" to scien- 5. Time dependent flow s l, ion animationtists and engineers who may not be completely familiarwith the inner workings of grid generators. flow solvers. Accomplishmentsand post-processors. Highly interactive computer pro- The first production version of TOPDUUG. The Onlygrams which provide graphical feedback have proven to Package for Design Using Unstructured Grids, has beenbe great tools for use in research. This effort focused produced. TOPDUUG is a highly interactive 2D un-on the -eneration of a computer program designed to structured grid generator/pre/post-processor designedintegrate several of the tasks in the CFD process into for ease of use and utility. TOPDUUG is currently be-an interactive, graphical package which was easily used ing used in the Aeromechanics Division for several en-by any engineer or scientist, not just by CFD code de- gineering applications as well as for code development.velopers. Future PlansApproach Future plans include modification of the 2D unstruc-Several different computer codes were being used to per- tured grid generation module to be more robust andform the tasks of grid generation and pre- ant, post- user friendly. Several scalar and vector functions will beprocessing within the Aeromechanics Division. These a(led to the post-processing module to increase func-codes were combined in TOPDUUG to create a general tionalitv. A user's manual describing operation of the"cradle-to-grave" tool for use in 2D CFD analysis. Inter- software will be written.active menus and windows were added to increase func- Publicationstionality and utility, An interactive on-line help system Gridley. M. C.. "TOPDUUG - The Only Package for De-was incorporated. Capabilities of TOPDUUG include: sign Using Unstructured Grids- User's Manual.' to be1. Fully interactive boundary condition specification published.2. Two-dimensional unstructured grid generation.including viscous grids

Several Screens Showing TOPDU)G Functions

7

Numerical Simulation of Turbulent Cylinder Juncture FlowfieldsDonald P. RizzettaCFD Research Section

Research Objective ber terms. For supersonic flow. the turbulence equationsThis work investigates the ability of an advanced two- incorporate a compressibility correction.equation turbulence model to properly simulate flow- Accomplishmentsfields which exhibit, complex three-dimensional separa- An existing vectorized three-dimensional implicit com-tions. Steady high Reynolds number subsonic and su- An codeti as eorifed thrdensionalcompdct co-- - ' ' upr~i~puter code was modified in order to accommodate so-personic flows past a cylinder which is mounted upright lution of the turbulence equations, and the code wason a flat plate are considered. These flowfields are char- validated. Both subsonic and supersonic cylinder'plateacterized by formation of a horseshoe vortex system thatwraps around the cylinder, and are of fundamental in- juncture flowfields were generated numerically. All corn-terest because of the intricate fluid phenomena which imonly observed physical features of the flows were repro-occur.e Inaddition, the intriate fid representat wih of duced by the calculations. A grid mesh step-sizt studyoccur. In addition, the configuration is representative ofresolution requireentsa variety of practical situations including wing/fuselage of tesions, and th effect ofsthetompresiits" ~of the solutions. and the effect of the compressibilityand wing/pylon intersections, struts of aerospace vehi-cles, submarine connino towers. turbomachinery. wind creto a xmnd eut ftecmuainehave been compared to experimental data in terms oftunnel model supports. architectural aerodynamics, and static and total pressure coefficients, velocity profiles.meteorological and geological applications, and surface limiting streamline patterns.

Approach Future PlansCylinder/plate juncture flowfields are sinuiated nu- The turbulence model equations will be used to simulatemericallv by solution of the time-dependent three- the flow over a delta wing performing a roll maneuver,dimensional compressible mass-averaged Navier-Stokes as well as the explosive rupture of an aircraft forebody.equations. which are integrated in time to achieve the Publicationssteadv state. Effects of fine scale turbulence are repre- Rizzetta. D P "Numerical Simulation of Turbulentsented by a two-equation(k - ) closure model which in- Cylinder Juncture Flowflelds." AIAA Paper 93-303,cludes a generalized formulation and low-Reynolds num- CJinlv 1993.

M f=2.5Re=735,33

SI -- .8 cm ,

S-- m1

Supters,.,i," Cylinder/Plate Junoture Flowfield

Turbulence Modeling for Thrust Reverser Flow Prediction MethodsKenneth E. WurtzlerInterdisciplinary and Applied CFD Section

Research ObjectiveThrust vectoring is one means to improving the ma- This data was analyzed to identify the dominant phys-neuverability and take-off and landing performance of ical processes which overn turbulence in these flows.aircraft, However. thrust vectoring, especially in close Modifications to the k-e turL•, Thnce model were basedproximity to the ground. can significantly increase the on these findings. The new coefficients in the resultingcomplexity of the flowfield around the aircraft. Prob- modified turbulence model were determined via numer-lems that can result include hot gas ingestion. reduced ical optimization for benchmark jet flows.lift, and foreign object damage to engines. To achievegood aircraft designs, it is necessary to have an accurate Accomplishmentsflow prediction method capable of resolving turbulence. Streamline curvature and vortex stretching were identi-This work addresses the issue of turbulence modeling fled as the two most significant influences on turbulence.for the inclined impinging jets associated with the use Modifications which account for these two features wereof thurst reversing in ground effect. added to the standard k-e turbulence model. In the

cases considered. the modified model consistently im-Approach proved the accuracy for the computed location of theThis effort combined numerical research with experi- ground vortex. Errors in the computed vortex locationmental results in order to produce a validated turbu- were as large as .50% in calculations with the standardlence model. The cases studied were an impinging jet turbulence model and could be of either sign. The mod-issuing from a nozzle normal to the ground plane and ified model reduced these errors to the range of 13-18%.one issuing from a rozzle inclined 45' into the cross-flow- Nielsen Engineering and Research. Inc. (Dr. Publications

Rat pWL-TR-92-3121. "Turbulence Modeling for Thrust Re-Robertverser Flow Field Prediction Methods". December 1992.eddy simulations (LES) to obtain mean flow and tur-

bulence data. Basic physical experiments, performed at Childs. R. E. and Rodman, L C.. "Turbulence ModelingStanford Unhiversity. provided additional flow data. for Impinging Jet Flows." AIAA-92-2672.

Normal impinging jet. with Mach=1.0. exiting from a pipe into low-speed crossflow.

9

The Effect of Leading-Edge Cross-Sectional Geometry on Vortex FlowAerodynamicsT. Sean Tavares - NRC AssociateCFD Research Section istic of high Reynolds number flows and to maintainResearch Objective good grid quality in the rest of the domain, withoutThe objective of this research is the detailed calculation using an excessive number of grid points.of viscous flow features which govern vortex separation Accomplishmentsfrom the leading edges of highly swept wings at high The figure illustrates a calculation of the flow about aReynolds numbers. The extent to which leading-edge clipped delta wing with 650 leading-edge sweep and avortex separation occurs can have a strong effect on lift, sharp leading edge. The flow conditions are 10' an-drag. and pitching moment. and depends in part on the gle of attack at Mach 0.4 and a Reynolds number ofleading-edge cross-sectional geometry. For wings with 9 x 106. Good agreement between the calculated andsharp leading edges, the point of primary vortex sepa- measured spanwise pressure distributions, as shown forration is fixed at the leading edge. However, for wings the station at 60% of the root chord is indicative ofhaving rounded leading edges. this is not necessarily the the successful calculation of both the primary and sec-case. and the flow may remain attached along part or ondarv vortex structures. These calculations representall of the leading edge depending on angle of attack. the highest Reynolds number turbulent flow cases runReynolds number, and Mach number. to date with the in-house Navier-Stokes code. In addi-Approach tion. comparisons of preliminary solutic-'w ,h•ained onThe approach taken is the computational solution of the coarse grids for both sharp and round-edged wings. re-3-D Reynolds-Averaged Navier-Stokes equations. They produce observed dependences of vortex flow patternsare solved using the implicit, approximately factored on leading-edge geometry.Beam-Warming algorithm as implemented within an in- Fiture Planshouse computer code (FDL3DI). Turbulence closure is Calculations are underway on the 65' clipped delta withachieved with the Baldwin-Lomax algebraic turbulence rounded leading edges. for which experimental measure-model. which has been modified to account for the ef- ments show separation along only part of the leadingfects of large scale crossflow separation. A special chal- edge at 100 angle of attack. Calculations are plannedlenge of this problem is to generate a grid with sufficient for both geometries on refined computational meshesresolution to resolve the thin boundary layers character- having 1.6 million grid points.

Press 0s1rfbjon at 60% Chowd

CWeIS Exp.

Cp 4

0.0-

015-

0'. o.N 0.6 o. .o

60% Chord

Vorticity

-25.00 0.00 25.00

( ;i,'il;~tj¢ "v'Ori,'x ll(,w rii'): )i~r,' ;•,,.v, .slarj,.,l d, 'elta W'1111 au',l pu.re- , dst-list 'i•lit l ") 1 ;I"UL"'

10

Vortex Breakdown Above a Pitching Delta WingMiguel R. VisbalCFD Research Section

Research Objective The angular delay and onset of breakdown are foundVortex breakdown represents one of the unresolved is- to be strongly linked to the pressure gradient prevailingsues in high-angle-of-attack aerodynamics. Further un- along the vortex axis, A description of the 3-D instan-derstanding of this complex fluid dynamics phenomenon taneous structure of vortex breakdown is provided foris required for enhancing aircraft maneuverability and the first time using critical-point theory. The region offor alleviating undesirable fluid/structure interactions reversed flow in the vortex core is associated with thesuch ,as tail buffet. The purpose of this research is appearance of pairs of opposite 3-D spiral/saddle crit-study the transient behavior and 3-D flow structure of ical points. During its early stages. the vortex break-the leading-edge vortex breakdown above a delta wing down is fairly axisymmetric. However, as breakdownwhich is pitched to a high angle of attack. proceeds upstream. asymmetric effects become increas-Approach ingly important and lead eventually to the formation of

a breakdown bubble. This bubble structure is open andThe flows are simulated by" solving the unsteady three- contains within itself a pair of stagnation points whichdimensional full Navier-Stokes equations expressed in are diametrically opposed and which rotate in the samestrong conservation law form. A general time-dependent sense as the base flow. A representation of the break-coordinate transformation is incorporated to treat the down bubble using an iso-surface of constant total pres-prescribed wing motion. The discretized equations are sure shows a great deal of resemblance to experimentalsolved employing the implicit approximate-factorization flow visualizations of axisymnetric vortex breakdown inBeam-Warming algorithm. The accuracy and stability a tube.of the scheme are enhanced by means of a subiteration Future Plansprocedure. Further investigation of the relation between the dif-Accomplishments ferent representations of the breakdown region (i.e. in-Calculations were performed for a 75' sweep delta wing stantaneous flow structure. streakline visualizations andpitching at a constant non-dimensional pitch rate Q+ = mean-flow measurements) will be pursued for unsteady0.3. from an initial angle of attack ai = 25' to a final breakdown above a stationary delta wing. Investigationangle Ci 500. The pitch axis was located at the wing of the parametric effects of pitch rate and pitch-axis lo-trailing edge, An assessment of the effects of numerical cation. as well as. compressibility effects is in progress.resolution and the favorable comparison with available Publicationsexperimental data indicate the computational approach Visbal. M. R.. "Structure of Vortex Breakdown on acaptures the basic dynamics of the transient breakdown. Pitching Delta Wing." AIAA Paper 93-0434. January

1993.

Computed bubble-type breakdown above pitching delta wing

11

Numerical Simulation of Delta-Wing RollRaymond E. GordnierCFD Research Section

Research Objective numerical scheme before computing the dynamic rollThe combat requirements for modern fighter aircraft ne- maneuver. The results showed excellent qualitativecessitate that they be highly maneuverable and able to and quantitative agreement with experimental measure-operate at high angles-of-attack. These requirements ments. Then a constant roll rate. 0 = 0.1325, "roll-and-lead to an interest in understanding the aerodynam- hold" maneuver from o = 0* to o = 450 was computed.ics of delta-wing maneuvers, including hi h-rate rolling The dynamics of the vortices during the roll maneuvermotion. The focus of the present research effort is the have been described, A lag in the upward leading-edgenumerical simulation of a constant roll-rate, "roll-and- vortex similar to that observed for delta wing rock washold" maneuver. The main objective is to describe the noted. The effect of the dynamic motion on the vortexdynamical behavior of the vortex position and strength, strengths has also been described. A simple, quasi-staticas well as the corresponding effect on surface pressure, explanation of this vortex behavior based on effectivelift and roll moment. angle-of-attack and sideslip angle during roll has beenApproach proposed.The governing equations solved are the unsteady, three- Future Plansdimensional, full Navier-Stokes equations in strong con- Current work on this project consists of the investigationservation form. A general time-dependent coordinate of parametric effects on the aerodynamics of the roll ma-transformation is introduced to allow for the body mo- neuver. The effects of roll rate. Reynolds number, andtion. The equations are solved using the implicit, ap- asymmetric initial conditions (i.e.. starting from nonzeroproximately factored Beam-Warming algorithm. Both a roll angles) are being studied. Calculations of turbulentfull block tridiagonal and diagonalized inversion schemes flows over delta wings using a two equation k-E turbu-are available. A subiteration procedure is also incorpo- lence model are also planned. The ultimate objectiverated to improve the accuracy and robustness of the of this work is to be able to numerically simulate delta-algorithm, wing rock both with and without vortex breakdown.Accomplishments PublicationsTwo fixed-roll angle cases (o = 0' and o = 450) Gordnier. R. E. and Visbal. M. R.. "Numerical simula-were computed to demonstrate the capabilities of the tion of Delta-Wing Roll." AIAA-93-0554. January, 1993.

Figure 12. Vortex Dynamics during a Roll Nlaneuver:X/L = 0.9

12

Nonequilibriumn Hypersonic Flowfield Prediction forFlow Past Blunt Bodies

CVD I) l'art-ji Sect iolt

Re'searchl Objvetive t otllctxvgei andt lit geitn nlolectiliar Oy~goil and n iI rogo t'lt nitr1 i c oxide tlte ctllll( jal ~ource teritis* lo itprorl'Ori nitiititric; modlo(!!iig of' tilt. foeii l'siig art, i'tivtd fromt Il Lt-Iaw of \Iaiss Act iot I he In11-

I lit' Navi.'r-Stokes eqiial ots withl itlltt' rate t'it'llical ki- pttrtaltict of a Iultilitonlttti ratulrt, titode I Ir the lproitrnett 1C a;jppitcahit I,) h Ighl-Wlem pt'ritutlre ailr [or iixpersoiiIC treatilileit of tile xtitrationial relaxat oll ill tlt- lprei(iflows lil cieltionfal and thieritial iioneqtiiblhritini. I Ioll of i til soirfact' heat I rallsfer is sihown 'Iillt varlat lQll

- Appjroadh III coilliplted shtock-st andoff dlistance suibsuint iates t I ital

Illt ;lxisynietlltt' Navier-Stoke.. I'iiautions. 5 spocl thle Nlacii numilbt'i jindependenlce prinlciple applicablet to('011 iiiit t(ltatoih. i vbrt iiii eiegv qutioelt'i gases does not appily to dissocl.at I itg fos h(-Olillliý eualons 3 ibatinalone,*-equtios, conttpuited st agntationl poit. t heat flux and lth surfact'aitt a tot~ai eiitrgy t'tjlaniol are soived Ill a 1`ilNlcilvti- le1 alirsi~tXelltcniaiolwt h '~

pled explicit llelt' lioti uslltý f lit, Roe, Viix-lDifferelii'e Split ht;ileat Ianir sitaowt exrllett roi mpeali ~triso %it'iith exi-'Schelp peIina dat itt'.dfon etli xernvlt olA ccomIplishmetsl(ll ducte~td Ii a shtock t in el antI front a set of earlier shock

tube e xper imn'it s-liT'e extenlsioin of thle Roe scitettie to inluilde tilt,' 1,1t01e Maxt im Plaitsl'ate, chemical kint'tic t'tqiatiolis follows the approac IEffects of ionliz/ationl aint radliatioll b'coimt Imtport ant atof i irossillal ando ('inneila. Secondl ordtr: acctirac, i ihrtmeauesadtes tb suidfrpoeachieved withI thle NiIi'S( ' approach. 'Ille full Navier- hge eleatrsadtlywl esu~dfrpoeStokes eqilatilols are solved with finite rate cltenii~strt, prediction of the flowfield and surface heat transfer oilfor thet flow past all axl-svltlmetrnc blunt. body at zero theblwcast.io lnts oisincidence at several Macit nutmbers ranging from Mach Pbiain10) to 18, at, altitudes of 22 kill and 37 km., Tile wall Josyula, Eswar and Shang, Joseph "Computa-temperature is set at 555 K and the maximum shtock tioni of Nonequilibrium Hypersonic Flowfields Pasttemp~erature is aroiuind 15.000 K. Both non-catalytic aild Hem isphere-Cyli nders." Journal of Thermophysics andfully catalytic wall boundary conditiotns for tite species Heat. Tr-alsfer, to be pilhiisllea in July 1993.are, utilized-

44.0 * 01__________________ Pt =0.0023P&

39.0- T,,j = 555.5K

* 1It.~.ModelI = =0.0127mn-. 34.0 ýý ý4111111`1tat tion (Model 3)

R.. Model 2 = 0.0234m

29-0o 11, Model 3 =0.0508m.,r ~lipeiaton(Model 2) Model 1. Thermocouple Gauge

24.0gnt~ Thin Filmi Gauge

10NModel 2. Thermocouple Gauge@019 .0 -T hin F ilm n G au tge *

14.0 " Model 3. Thermocouple Gauge 8

90 ->-Thin Film Gauge a

Conuputatmo,, (Model 1) ~

4.0 W0.00 0.05 0 10 0.15 0.20 0 .25 0.30 0.35 0.40

Distance (in)

Comparison of surf. ce heat transfer distribution with shock-

tunnel experiments for three d iffere nt- sized models

13

Stability Analysis of a Combined Couette-Poiseuille, Two-Fluid FlowLt John J. NelsonInterdisciplinary and Applied CFD Section

Research ObjectiveIn this work, the complete theory is developed for the order about the basic flow. Systems governin* the dis-Iinear stability analysis, energy analysis and weakly non- tortion of the mean flow, the rise of the second ,armoniclinear analysis for a combined Couette-Poiseuille, two and growth rate of finite amplitude waves are found andfluid flow. The linear stability analysis determines the solved.conditions at which the basic, laminar flow will become Accomplishmentsunstable and transition occurs. The energy analysis de- This worktermines the mechanisms which cause the instability. cas wons co r pletely derives a theory and provides testIn multi-fluid incompressible flow, instability is influ- case solutions for transition prediction coding of incom-enced by five mechanisms: viscous dissipation, Reynolds pressible flows. The flow situations studied in the workstress, interfacial friction, density stratification and sur- are the standard test cases for linear stability code accu-face t:nsion. Weakly nonlinear analysis determines if racy. Our linear stability analysis is in agreement withstanding wave solutions are possible in the flow ensu- the standard answers. Going beyond linear analysis,ing after transistion, and if so. determines the transient this work is the first to identify the mechanisms leadingamplitude of the waves, to the instability. Specific questions relating to normal-

izations are also addressed for the 'erst time.Approach Future PlansThe linear stability analysis is accomplished by linearly The theory developed is being employed in an incom-perturbing the continuity, Navier-Stokes and interfacial pressible Navier-Stokes code and used to study a prob-equations. A separable solution is assumed, which leads lem of current interest. Also planned is the developmentto a complex eigensystem problem, which determines of a stability theory for compressible flows. This theorythe growth rate of infinitesimal disturbances in the flow. will be used to give transition prediction capability toFull viscous and surface tension effects are included, our in-house compressible Navier-Stokes code.The results of the linear analysis are used to derive aenertv anedysis. The weakly nonlinear analysis is ac-complished by expanding in a small parameter to third

1ow 1.11~

9W.,

w "o. S..so o ,.d .io

6. S

.4

20.1.03.0.607 4.76W10i03010.i*

k kc(a) (b)

(a) Linear stability analysis of two fluids in a Couette flow. R represents the Reynolds number and k the wavenumber. (b) Energy analysis for the Hlow of (a) at H? = 6300, The term "body tern** represents viscous dissipationplus Reynolds stress.

14

Computational Aerodynamic Analysis of a Decoy ConfigurationFrank C. Witzeman JrInterdisciplinary and A.,lied CFD Section

Research ObjectiveAerodynamic analyses of air vehicles operating over a results were compared with the semi-empirical methodrange of flight conditions can be performed by a va- results and experimental data tintegrated onlyl. Theriery of methods including analytical. semi-empirical, final solution matrix consisted of one roll angle, threeComputational Fluid Dynamics (CFD). and experimen- Mach numbers. and several angles-of-attack.tal testing. Limitations in both the semi-empirical and Accomplishmentstesting techniques for a decoy configuration lead to the Euler solutions for the decoy at Mach numbers of 0.5use of CFD to fill gaps in the required data. Specif- and 0.7 have been obtained for several angles-of-attackically. aerodynamic data at transonic, high angle-of- and 0' roll angle. The low-speed results compare wellattack conditions were needed. Primary interest was to the experimental data up to intermediate angles-of-in obtaining in-plane (pitch) forces and moments due to attack (200). whereas the Mach-0., results show discrep-inertial effects (inviscid flow). Effects due to roll angle ancies beyond 100 angle-of-attack. Compressibility ef-and small changes in Mach number and angle-of-attack fects have a major impact on the aerodynamics as- notedwere of interest. but secondary- in the CFD results. These results also indicated thatApproach the semi-empirical method did not accurately predict

cent er-of-pressure location, especially for the aft-bodyThe in-house Euler code MERCURY was used to obtain component.CFD solutions for the decoy. A structured grid consist-ing of 45 blocks and approximately 720.000 grid points Future Planswas generated from the GRIDGEN code. The block- Additional CFD solutions at Mach-I.2 and angles-of-ing schenie was beneficial in reducing computer niem- attack up to 900 are planned. These results are lackin,_,:ory requirements where the maximum memory required in the experiments. All results will be tabulated fordepends only on the largest, block. Integrated and corn- the total integrated and component forces and moments.ponent forces and moments were compiled for each so- Compressibility effects at transonic conditions will belution. The components consisted of the nose. center- investigated in more detail with the CFD solutions.body'. aft-body, and tail. The integrated and component

S... ;::M =0.7

ALPHA 0 dog

Mach Niiiber C.onrours ind V ~loiry \'t,;,rs for Dpt'() •"ofihliiratoin

15

Euler Analysis for Delta Wing DesignLt John S. SeoInterdisciplinary and Applied CFD Section

Research Objective AccomplishmentsAt. the request of ASC./XR. the feasibility of a 14% thick Over 50 solutions with Mach number ranging from 0.5flyina, delta wing to sustain efficient flight at a design to 0.95 and angle-of-attack ranging from 0 to 15 de-speed of Mach 0.90 and lift coefficient of 0.60 was tested. grees were computed on two of AFMC's supercomput-CFD analysis was applied for this configuration as the ers: Wright-Patterson AFB's Cray/XMP and Eglincalculation of wave drag using empirical methods were AFB's Cray/YMP. It was found that at Mach = 0,90.unreliable. This delta wing has a leading-edge sweep and even at a lower Mach of 0.80. this 147 thick deltaangle of 53-1 degrees and half-wing planform area of wing will not achieve the desired C, of 0.60. A strong525.87 square feet. shock slightly past the leading edge, and reversed flowApproach near the wing tip contribute to the C0 loss.Since the parameter of interest was wave drag, an Euler Significancesolution was sufficient. for this analysis. Three cross- Existing CFD tools allow WL/FIMC to perform an Eu-sectional cuts of the wing were prepared in an interac- ler study for this relatively simple configuration withtive graphics grid, generator. 13G/VIR.GO. to create a ease. This particular study lasted 5 weeks from begin-full three-dimensional geometric surface definition. A ning to end.single C-H type grid with dimension of 61x40x70 was Presentationscreated using another interative multiple-block grid gen- The results of this analysis were presented at theerator. GRIDGEN. MERCURY. an in-house 3-D Euler Dayton-Cincinnati AIAA mini-symposium, 25 Mar 93.flow solver based on Jameson's algorithm, was used forwave drag calculations.

MEFRCURY solution or delta wing at Mach 0.80 and 6 degrees angle of attack

16

Numerical Simulation of a Jet Expulsion from an Aircraft Forebody,With Comparison to ExperimentCapt James Muqdy - CFD Research SectionResearch Objective grid points). without the jet blowing. Agreement be-This research investigates the effect a sonic, underex- tween the numerical solution and experimental data ispanded jet issuing from a generic aircraft forebody. has excellent. The solution for the jet-on case is still onlyon the airflow around the aircraft, and on the resulting partially converged, though indications are that the au-aerodynamic forces, merical solution will agree closely with the experimentalApproach data.An implicit. Beam-Warming finite difference scheme is Future Plansbeing used to solve the Navier-Stokes equations around The jet-off case will be refined using a finer grid (ap-the aircraft forebody. A k- - e turbulence model is used proximately 1.25 million grid points) to provide valida-to account for the effects of turbulence. The result- tion for the CFD scheme. Once this is done, the genericing numerical solutions are compared with pressure data aircraft forebodv will be modeled to conform tQ typicaland flow visualization video and still photography taken flight, characteristics of a jumbo-jet. The free-flight casefrom wind-tunnel tests. Pressure data is in two for. ý. will model flight at Mach 0.75. at an altitude of 30.000the first being pressure tap data. the second being data feet.taken from a new'pressure sensitive paint technique. PublicationsAccomplishments Melville. Reid. "Numerical Simulation of a Sonic. Un-The three-dimensional Navier-Stokes equations. with a derexpanded Jet from an Aircraft Forebody'" FAA Air-k - c turbulence model have been solved around the craft Hardening and Survivability Symposium. Augustgeneric aircraft forebody on a 108x88x43 grid (408,672 1992 Atlantic City NJ

Re = 10.

Streamlines Aromid a Jet Issuing from a Fuselage

17

Numerical Analysis of a Chined Forebody with Asymmetric StrakesKenneth E. WurtzlerInterdisciplinary and Applied CFD Section

Research ObjectiveVortex forebody control (VFC) is increasingly being Convergence was determined by a drop of four orders ofviewed as a way to provide greater directional control magnitude of the residuals and leveling off of the bodyof fighter aircraft at high angle of attack. The major forces.objective of this work to examine the effectiveness ofvarious strake sizes on the yawing and rolling moments Accomplishmentson the Fighter Lift and Control (FLAC) forebody alone. Solutions were obtained at 15.25, 35. and 450 angle ofA secondary purpose is to validate the use of an Euler attack at a Mach number of 0.18 Initial calculationscode for the prediction of flow over a chined forebody at on the clean forebodv resulted in no side force beinghigh angles of attack. produced. which is theoretically correct for Euler codes.

The results showed that a short, narrow strake producedApproach a yawing force that acts in the direction of the strakeThe geometry utilized was the full FLAC forebody with side. Wider strakes pushed the vortex away from the

a aforebodv, creating a yawing force that acts in the direc-antion of the other side. The strong vortex that separatescut. Different sized strakes were attached to the left captred wel the Eerside at the apex of the nose. A 1.7 million point, multi- from the chined forebody is captured well by the Eulerblock grid was constructed using I3GVIRGO. In order code,to simplify the grid generation and to provide for quick Future Planssubstitution of different strake sizes. the ability to create ivarousstrke ize wa bult ntothegri, Gid pacng Numerical analysis of normal and tangential forebodyvarious strake sizes was built into the grid. Grid spacing blowinig will be'simulated and compared to thf, asvru-_normal and tangential to the surface was kept equal soas to produce a dense, equally spaced gridded domain metric strake results. Additionally. an unstructured gridcapable of capturing the core of the vortex. The flow approach will be used to compare Euler and Navier-past the forebody was computed by using an in-house Stokes results for strake and blowing applications.Euler code. MERCURY. The full forebody was mod- Publicationseled so as to capture any asymmetry. Convergence on ,urtzler K. E. -Numerical Analysis of a Chined Fore-the forebody alone case was reached in about 2.000 it- \Vurth KE.Nmeric Analss o aper 93-erations. Each strake case needed approximately 3.000 body with Asymmetric Strokes. AIAA Paper 93-0051.more iterations. January 1993.

Vorrviiv contours awd particlees nvr FLAC for'o'ot with lar and srnallest straki at 45' an,,le cfattak

• . . , , n II

Inviscid Hypersonic Flow Around a Conically-Derived WaveriderCapt Gregory 0. SteckleinInterdisciplinary and Applied CFD Section

Research Objective The on-design solutions were then validated against theThe current high level of interest in hypersonic flight, analytical results which were based on hypersonic smallbrought about to a large extent by programs such as the disturbance theory. HSDT. and an integral momentumNational Aerospace Plane. has focused much attention development. Quantitative results matched well withon the stringent requirements of aerodynamic design at the analytic results to within 535 for lift and drag cc-these demanding flight conditions. The major objective efficients. Off-design calculations were run for the wa-of this research is to analyze the inviscidly dominated verider at Macl 10 for an AoA range of ± and atflowfield of a waverider forebodv for on- and off-design 0' AoA and a Mach range of 6 to 15. For the inviscidflight regimes. calculations, the off-design perturba'ions were charac-Approach terized by wave drag effects. Of particular interest wasThe governing equations are the three-dimensional Eu- the linear wave drag bucket associated with perturba-ier equations utilizing a finite-volume solver using spa- tions to the on-design attitude. For the range of e.tti-ial Roe-averaged. flux-difference splitting They are tudes investigated in this research. wa-ve drag increased

solved with a standard McCormack explicit time inte- approximately 107 per degree angle-of-attack.gration schenit. Convergence is accelerated at the ex- Future Planspens., of time accuracy by the local time-stepping tech- Current -lans are to reevaluate this waverider configu-nimie ration with a mass-addition leading edge. Thle implicit,Accomplishments full Navier-Stokes mode of the current algorithm willThree general classes of solutions were investigated for also be employed to investigate the effects of shock wavetwo levels of grid refinement. Inviscid solutions were and boundary layer interaction.o)btained for on-design conditions of Mach 10. 30.5 km Publicationspressur, altitude and zero degrees angle-of-attack (AoAi Stecklein. G. 0.. -Numerical Solution of Inviscid Hv-and two off-desian flight regimes. Qualitative results il- personic Flow Around a Conically-Deri ed Waverider.lustrated excellent shock capturing and flowfield forma- AIAA Paper 93-0320. ,anuary 1993.tion characteristics with grid refinement.

in\isid Macb Contours I.\.F. = 10•U ' F I)

IOU.,

The Flowfield Past the X24C Reentry VehicleDatta Ciaitonie - Ui nIversal Energy% Systemis IncCFD Research SectionResearch Objective past the X24(i - l-101 configuratlion is then wove wIth aThe purpose of this research is to examine the flowfield mnesh of r(oughl% half million grid points. The io-calledabout complete lifting-bod% configurationis at hyper- -carbuncle" anomaly in the nose re-ion is eliminated bvsonic speeds utilizino an accurate and efficient numeri- a combination of difft-rent splittings without affectitngcal algorithmi. The ability of a hig-h resolution schemet tc thp: overall accuracy. Analysis of the computed turbu-provi'de accurate engineer .ing estimate,, of surface ther- lent flowfield show good agreement with exp-rimentalmal and nwe(hanical -loading is examined, surface pres~surre and heat tranisfer data. Several aspects

of the computed laminar and turbulent flowfi#lds are- in-Approach VeStIgared. including Ih ineatono h ortex struc-The full 3-D compressible Navier-Stok-es equations in turte with the rail finsmass-averaged variables are solved in a finite volume for- Future Plansmulation. Roe's flux-difference split method is used for The theoretical rnodelI'will be extendeýd to incorporattt~he inviscid fluxes while viscous fluxes are centered.- T he'Iimplicit numerical algorithmi is based on Gauss-Spidel hii~h temperature effects such as thermo-chemical non-line relaxatin Turbulence closure is achieved withi the equi'librium and ionization A 2-D version of the code.Baldwin-Lomnax algebraic eddy viscosity model, employing nion-equilibriumi five-sp-cies chemistry and a

sipeharmonic oscillator model for vibratioiia. neAccomlishmnts xciTtaion. has already be en developed

The choic,ý of algorithm is determined with a crit- Publicationsical examination of NMUSCL based higher order up- 'Numerical Simulation of Flow Past tie X24C Reentr%wind methods andi a range of limi-ters. Spe-IA 33l<1tclfi-c eaiphasis is placed on the prediction of Vehicle." D. cGaitonde and J. Shani. AIA 301.3shepat transfer ratef. in viscous flow-, The flow Aerospace Scic nc:s Meeting &- Exhibit. Reýno 19621.

LLaminar Turbulent

('ni~i 8.' -i 6-. 1 af--t i.'m X? It' wr-frrý vo l.k

20

Deniscp ,atin Over a Swep W ing with a in Ai

Interdisci~j J~ nskzj a d1 C

Inary and" "apt 1F-0.laIReseA c O b j ci -b

e~ walj

D S cini

h E th alpyTap In th S ' g t the affect of a ead ing edge

atneoo~ ej e p tea lSt

a

surface h a Irn sof a Swept~ex,Appra et tanser at ~,wing" 1 On2 the rnmal surlace ober~

The IIYLDA cod e (Hv ers

oLokDe a td the f wie I P t hean I gaplfWa,ý used xI ý, Clio-Stagnatio IIga- -

in hi i (ivpero sPt n tr I n the

,s aviet

re,-i nm O

rs t hle u gap a fl t ha P lane sho ~l t f at)u nal is)of' the usep r ea n at upý fo se~ ra

ge ea l .se pnt, el th full duVortexh

upat~e a s -Ig I'hcha Inln asrtýýfull V_ oupld -ý'"ýPn~ on910n ntear the 'gh hea~ rt e

ul witthIng due

mnpe Of rac t Y Oft efl c aton, tei he i -~ . F' 0 vortfe x.a dwi e h a.

Thatinose co a) t ied t orhe at fd re re n hepi, Pr wea ll,

th owfleld cac ai~ withe Thie re 'le

agawept

the gapo r e , nI.s 5 i ssipao f s eat

Ac o e~ .Cgridat

dens-e

eaiout e o 'h) e ac l ne d h iasngiI a te Il eaFtungrtes aresecio wih ona gidsft ensý ndheled Thilsud cndit t o f d t he leadi e~h

Theun thefi a u g ap swp e pto ain setion was c ahi

in~~ Cea

.

rep es nt ag Sed e

atnn ondt o Sh~ ' e Man h w an fra dh.ng s cto fo t ghe W ed Ani, , ea IAA, aie

eqCap , 1 or~ bighec th nhat t rase r Aug,~ lýn ` A'raIu tedv withu

0a thW nlrrznle w a Pod

to j il b c

An h a p

S.j 3 3

l h Sp e i a nt n e

*'%CCom ji~hr, vo Ia op,,s nae ihi PornT, -gP,:oaprorate e alPs il I

Theion an te'- Frt a~ fr '7rt"nt21t

Bomb Blast SimulationWillam Z. StrangInterdisciplinary and Applied CFD Section

Research ObjectiveSimulate air blast loads induced by the detonation of a inertial forces involved. viscous effects were not simu.-small charge. lated. Finally. radiation effects were assumed negligible

Approach due to the short time scale involved.

The general approach taken was to incorporate a bomb Accomplishmentsmodel into an existing computational fluid dynamics One of the Aircraft Loading Tests conducted by(CFD) code. COBALT was chosen as the CFD code to WL/FIV was simulated In the test. a bare sphericalhost the bomb model. This code. an unstructured grid. charge was located at the centroid of a rigid, right-higher-order Riemann solver, was chosen because of its circular cylinder with semi-oblate spheroidal endcaps.robustness and the ease with which the bomb and its The particular test. simulated was chosen for its accuratesurroundings could be discretized. The bomb model is pressure histories at several recording locations Thevery simple owing to the assumption that the detailed simulation covered the first I I milliseconds of the blastphysics of charge detonation are unimportant at dis- event. The accompanying figure depicts the pressuretances large compared to the charge radius. A corollary contours within the cylinder 1.56 milliseconds after det-to this assumption is that the mass of the detonation onation. In it. the primary blast wave, first sidewallby-products is small with respect to that of the sur- reflection and implosion shock are seen Comparisonsrounding medit m. With these assumptions. a charge between experimental and numerical pressure historiesmay be simply modelled as a mass and energy source. at various recording locations show excellent agreement.The mass and energy transfer rate is assumed constant Future Plansand is calculated from tabulated detonation velocities We expect tosimulateadetonation within a rigid. three-for the charge material at hand. Owing to the high densio selage.

dimensional fuselage.

Press.ur,, Conrtors "it 1.56 millisecondls ifter Charge [D)tounaon

Parallel Processing for Computational Fluid DynamicsStephen ScherrCFD Research Section

Research Objective the Intel Touchstone Delta massively parallel computer.This effort describes the parallel structure of an explicit The parallel design features a 2-D pencilled block dataNavier-Stokes algorithm and its implementation on a decomposition with only nearest neighbor communica-massively parallel mnessage-passing computer. tions. This is well-suited to the two-dimensional mesh

architecture of the Delta machine.Approach As a test of the implementation. the vortical flow aroundThe analysis of complex fluid flows and the future cou- a 750 swept delta wing was simulated numerically forpling of CFD techniques in interdisciplinary applica- flow conditions of a = 30'. M.. = 1.95. and a Reynoldstions require ever increasing amounts of computational number of 4.48 x 10. The computational grid was anpower. Trends in computer architecture indicate that 80 x 80 x 80 mesh with an H-H topology. Timing re-the required computational power will only be avail- sults for the main computational loop indicate a factorable through the use of many processors all working to- of 10 speedup over a single Cray-2 processor when 500+gether to perform the analysis. The architectures which processors of the Delta machine are utilized (see chartcurrently show the most promise are those in which below). The best pertormance occurs when the workeach processor has its own private memory and com- load is evenlv balanced across the processors. Variousmunicates with other processors by passing information methods for remapping a 13 x 39 set of blocks into athrough messages. Samples of these are tightly coupled 16 x 32 physical processor layout were examined. Themassively parallel computers, and loosely coupled net- l oworksbest laout occurs by maximizing the number of pr-To perform a computational fluid dynamics analysis in cessors having nearest neighbor communications. Thisparallel. the entire problem must be divided up into configuration is illustrated in the diagram below.pieces- which are then assigned to the available proces- Puture Planssors. For best performance. this decomposition should Future work in paralle' rocessing for computationalevenly distribute the work among the processors, and fluid dynamics will include implementation of a stan-minimize the number of messages which need to be dard turbulence model under a blocked data decompo-passed between processors. In addition. support mech- sition, extension of techniques to other algorithms, suchanisms such as input/output techniques should be de- as an explicit-implicit flux-split algorithm, and investi-veloped to take advantage of system features. gation of methods for parallelizing unstructured grids.Accomplishments PublicationsAn exist-ing serial implementation of the two-step ex- Scherr. S.J.. "Implementation of an Explicit Navier-plicit predictor-corrector MacCormack algorithm for Stokes Alorithm on a Distributed Memor9 Paralle

lution of the 3-D Navier-Stokes equations was ported to Computer. AIAA Paper 93-0063. 1993

35.

30-

25

20

4) * Cray-2tn 15 .. . . . . . . . . .. . . . . . . . .-• , ' --i ' -, -! --= ; ---' -" " -.... ------ -----

+ T•p II i ll

Reconfiguration for Load Balancin14

0 64 128 192 256 320 34 4 2

NodesT *nng results for FD_3D--EP

23

Advanced Technology Air Vehicle Geometry Data BaseHoward T. EmslevInterdisciplinary and Applied CFD Section

Research ObjectiveAccurate geometric representations of USAF aircraft To reach our goal. a data base structure. a set of geome-and weapons are necessary to perform many Air Force try formats. and the tools to store, retrieve and interac-support activities such as weapon and vehicle modifi- tively view complete air vehicle geometries are required.cations. Computational Fluid Dynamic (CFD) analysis, Concurrent with the development of the system. WL isand weapon fit checks. Acquiring necessary geometries colectin current and future SAF air vehicle geoe-is a difficult task due to: a lack of eometri tries to store in the data base. Data formats specified

ieometric standard- by CALS will be used to acquire the geometries, andizdatdon in the industry. the proliferation of computer- A MIL standard modifications or clarifications will be ad-aided design tools. and the absence of complete U'SAF dressed to reduce problems in translating geometric rep-aircraft external geometries. Standardization of geome- resentations from different systems.try format.s under the Initial Graphics Exchange Spec-ification (IGES), Product Data Exchange using STEP Accomplishments(PDES,. and Computer-aided Acquisition and Logistic A contract to develop the data base structure and inter-Support (CALS) rmay be the first step toward resclv- face software was issued to Rensselaer Design Researching the problem. but another issue involves the acquisi- C'nter in May 1992. A non-uniform rational b-splinetion and storage of air vehicle geometries for use by Air geometry format (STEP NURBS) was selected and theForce organizations. With accurate geometries in hand. preliminary interface software has been developed. Ad-the timeliness and ability to perform support activities ditional work in progress involves interfacing with ex-can be improved, duplication of effort can be reduced. isting standards, and the inclusion of topology betweenand common geometric configurations can be provided geometric entities. The former involves format conver-to Air Force research and support. organizations. sion of IGES entities into STEP NURBS entities forApproach storage and retrieval.

To perform support activities for Air Force systems. ge- Future Plansometries of USAF aircraft and weapons should be ac- As the contracted effort continues, air vehicle geometriesquired as advanced surface representations and stored will be collected for conversion and storage in the datain a data base. The goal of this program is to reduce base. In-house effortrs associated with this effort are con-the time required to perform CFD analyses by providing centrating on efficient ways of interfacing NURBS sur-accurate geometric representations to CFD users. faces with existing and developing grid generation tools

for CFD.

Advanoed Technology Air Vehirlo G;,mc, r, Dat Base Browser and \'ewer

('Qnriz(,nt .l tail data is displaypd In inhrr:,wr and highli ghted in viewer)

21

Computational ElectromagneticsJ. S. ShangFlight Dynamics Directorate

Research Objective AccomplishmentsThe ultimate objective of the present effort attempts The characteristic-based methods have been validatedto establish a numerical simulation capability of elec- by applying them to a three-dimensional wave guide andtromagnetic phenomena for air vehicle technology. The a singular electric dipole problem The present approachbasic process involves the transition of innovative tech- has demonstrated a unique potential in alleviating, andniques from computational fluid dynamics (CFD) to under some special circumstances in eliminating. thecomputational electromagnetics (CEM) for signature fundamental dilemma of reflecting waves from the trun-analysis and control. target identification, microwave cated spatial domain. The numerical pru.c-dures alsocommunication. as well as environmental hazard pro- exhibit significantly improved numerical stability prop-tection. Specific objectives are to first develop efficient erties and reduced alaising errors from the current corn-numerical procedures for solving time-domain, three- putational electromagnetic methods in use A very highdimensional Maxwell equations. and then map numeri- data processing rate of six gigaflops has also been at-cal algorithms onto massively parallel processors. tained by a present finite-difference procedure. MostApproach importantly. on a massively parallel system. the present

second-order accurate finite-difference procedure is scal-The time-dependent Maxwell equations for propagatin$. able. The scalability is sustained up to 512 nodes, withrefracting and Jiffracting eletromagnetic waves consti- only a 3 percent parallei efficiency degradation.tutes an initial value problem. The hyperbolic system ais solved by characteristic-based schemes, which are de- Future Plansrived from eigenvalue and eigenvector analysis. The Further research will be pursued to refine the finite-three-dimensional system of equations. when written in volume procedure for an air vehicle radar signature anal-flux vector form. can be di gonalized in each single di- ysis effort. A comparative study is also planned to as-mension. The concept of flux-difference splitting will sess the relative merits of the finite-volume and finite-permit windward schemes to mimic the wave mechanism difference approaches in application to complex scatterof information propagation. In three-dimensional appli- geometries, and numerical efficiency on massivtly par-cations, the cyclic fractional-step or the time-splitting allel computers. The characteristic concept will finallyformulation ensures a second-order accurate procedure be extended to develop a meshless computing procedurein time and space. Both finite-difference and finite- vol- for efficient and accurate CEM applications and diverseume solving procedures will be implemented on a gen- dual-use purposes.eral curvilinear coordinate system to accommodate corn- Publicationsplex scatter shapes. Extremely high numerical efficiency Shang. J. S.. "A Fractional-Step Method for Solving 3-for high frequency wave simulation are being achieved D Time Domain Maxwell Equations," AIAA paper 93-through a domain decomposition technique to minimize 0461. Jan. 1993.the data communication on a massively parallel proces-sor.

Field Intensitis for Ele,-trir Dipole

25

List of Authors

The mailing address for all authors is:WL/FIMC

ATTN: Author s name

2645 Fifth St., Ste. 7

Wright-Patterson AFB. OH 45433-7913

NAME PHONE NUMBER E-MAIL ADDRESS

Aftomis, Michael J. (513) 255-7127 aftomis-Sifim.wpafb.af.mil

Emsley, Howard T. (513) 255-4522 j emsleywfim.wpafb.af.n-ilFite. Phillip M. (513) 255-3413 fite9fim.wpafb.af.mil

Gaitonde, Datta (513) 255-7127 dattadfim.wpafb.af.mil

Gridley. Skip (513) 255-6207 cgridlev:Kfim.wpafb.af.mil

Gordnier, Raymond (513) 255-7127 gordni2fim.wpafb.af.mi jJosyula, Eswar (513) 255-2455 josvu.ai.im.wpafD.af.mi

Kinsey, Don W. (513) 255-3876 kinsevyfim.wpafb.af.mil

McClure. Thomas H. (513) 255-4522 mcclure •fim.wpafb.af.mil

Mrozinski, Denis P. (513) 255-3788 moringfim.wpafb.af.mil

Mundy, James A. (513) 255-7127 mundyvafim.wpafb.af.mil

Nelson. John (513) 255-3413 nelson'.Lfim.wpafb.af.mil

Rizzetta. Donald P. (513) 255-7127 rizzettaafim.wpafb.af.mil

Scherr, Stephen (513) 255-2455 scherrCfim.wpafb.af.mil

Seo. John S. (.513) 255-4522 seo@fim.wpafb.af.milShang, Joseph S. (513) 255-5616 shang'fim.wpafb.af.mil

Stecklein., Greg (513) 255-3761 steckdfim.wpafb.af.mil

Strang, William (513) 255-4522 strang'.fim.wpafb.af.mil

Tavares, T. Sean (513) 255-7127 tavares~fim.wpafb.af.miI

Visbal. Miguel (513) 255-2455 visbalffim.wpafb.af.mil

Witzeman. Frank (513) 255-3788 witzernan~fim.wpafb.af.nil PWurtzler, Kenneth E. (513) 255-3761 wurtzler4fim.wpafb.af.miI

DSN phone number is 785 - plus last -4 digits, listed.

26