Electromagnetic Launch Vehicle Fairing and Acoustic Blanket Model ofReceived Power using FEKO

Dawn H. Trout 1,2, James E. Stanley 2, and Parveen F. Wahid1

)School of Electrical Engineering and computer ScienceUniversity of Central Florida, Orlando, FL32816, USA

dawn.h.trout@k:nights.ucf.edu, wahid@mail.ucf.edu

2 Kennedy Space CenterKSC, FL 32899, USA

dawn.h. trout@nasa.gov

Abstract - Evaluating the impact of radiofrequency transmission in vehicle fairings isimportant to electromagnetically sensitivespacecraft. This study employs the multilevel fastmultipole method (MLFMM) from a commercialelectromagnetic tool, FEKO, to model the fairingelectromagnetic environment in the presence of aninternal transmitter with improved accuracy overindustry applied techniques. This fairing modelincludes material properties representative ofacoustic blanketing conunonly used in vehicles.Equivalent surface material models within FEKOwere successfully applied to simulate the test case.Finally, a simplified model is presented usingNicholson Ross Weir derived blanket materialproperties. These propeliies are implemented withthe coated metal option to reduce the model to onelayer within the accuracy of the O\;ginal threelayer simulation.

Index Terms - FEKO, MLFMM, NicholsonRoss Weir, Resonant Cavity.

I. I TRODUCTIOWith multiple contI;butions from the range

and surrounding radio frequency (rf) emitters,defining the electromagnetic environment forspacecraft can be a daunting task [I]. Determiningthe environment inside the vehicle fairing presentsfurther challenges as field distribution within thecavity is influenced by resonances which require afull wave solution to achieve the desired accuracy.

An added concern is that most spacecrafttransmitters are in the GHz frequency rangemaking the structures electrically large andmemory requirements a constraint for many of the3D electromagnetic simulation tools available.

In this paper, two cases are evaluated: a threelayer model, and a one layer model. The threelayer model of a vehicle fairing with layeredacoustic blanketing materials characterized by thinsurface approximations is first presented [2]. Forcomparison and validation purposes the test case[3] is summarized here and used as evaluationdata. Finally, an equivalent one-layer model isdeveloped using material properties predicted withS-parameters measurement and implemented in toFEKO standard coating option.

II. FAIRING FIXTUREA computational fluid dynamics fairing test

fixture was modified by lining the Lexan outershell with industry grade aluminum foil [4]. Thefairing has three sections bolted together and ametal frame outer suppOli structure. This fixtureis representative of typical launch vehicles,although at a smaller scale with a height of 2meters and a diameter of 0.6 meters. Thealuminum lined fairing fixture is shown in Fig. 1.Transmit and receive double ridge guide hornswere placed at the bottom and top of the fairingfixture, respectively [5].

Lining materials were added to the inside ofthe test fixture to simulate typical acousticblankets inside vehicle fairings. Kapton is

https://ntrs.nasa.gov/search.jsp?R=20110014945 2018-05-11T11:37:49+00:00Z

EI.ctJlcfl.ld [VIm)

8.2<1838

7."8525.01'.703782.88l.O7I.lICl0.18

Electric field !VIm)

8.398.477.648.816884783.832.811881060.13

distribution and power received at I GHz using asurface impedance of 0.0 15 ohms reveals excellentagreement.

commonly used in space applications for itsfavorable thermal insulating properties. DuPont'sKapton 160XC, designed to maintain a surfaceresistance of 377 ohms with inherent RFabsorption propelties, is utilized as the outerblanket layers while standard Y2 inch foam is usedas the internal layer.

The test results from this fairing fixture withacoustic blanketing are used for comparison withthe three layer and one layer models presentedhere. The goal is to obtain an equivalent one layermodel that has similar test data correlation as thethree layer model.

Fig. 2. Field distribution of aluminumfairing - MLFMM and MoM.

Similar results were found with FEKO's lossymetal feature which has minimal computationalpenalties compared to the efficiency of a perfectelectric conductor (PEe). FEKO evaluates theinput material propelties, such as permittivity andconductivity, to obtain a representative impedanceterm, Zs, which is then added to the standardelectric field integral equations used for PECstructures as in (1) [6,7].

Where,

Ei is the field due to an impressed source in theabsence of the scattererEs is the scattered fieldIs is the equivalent current density

Fig. I. Test fixture with CAD model.

III. THREE LAYER MODELA commercial computational electromagnetic

software tool, EM Software Systems, FEKO isutilized in this study. The multilevel fastmultipole method (MLFMM) feature isimplemented to extend the method of moments(MoM) technique to higher frequencies. AsMLFMM is an iterative technique that relies onconvergence to a specified error, accuracycomparisons were made. Figure 2 demonstratesthe adequacy of this approach for an aluminumcavity represented by an impedance sheet withboth MoM and MLFMM techniques. The field

Antenna pattern models presented in [2] of theEMCO 3115 hom developed within FEKO areimplemented in this simulation. Replacing thehom model with the hom pattern affords asignificant savings in computational resources.Parallelization of the FEKO code via pre­conditioners, such as the sparse approximateinverse, supports solutions for detailed electricallylarge structures as presented here [8].

A combined blanketing and composite fairingstructure model was presented in [9]. In thispaper, it is desired to first represent the layersseparately for direct test comparison. Fig. 3depicts the test fixture layers and the compositemodel within FEKO.

0:·1

Olll.

-1~I'h'J"1(

-·J-l,·nS!<'

lJtl .'90 l.llO 110

LEO'

11<11,.

'''''I

"Three Layer Model to Test Data Comparison

Fig. 4. Acoustic blanket model to testcomparison.

A. Methodology selectionTruncation of the scalar Green's functionimplemented with the addition theory series inMLFMM introduces an error that can becontrolled in open structures, but difficult toachieve sufficiently accurate results in electricallylarge reflective cavities [I 0]. This residual errorcan, in effect, numerically excite the cavity. Thus,convergence is improved by using layerrepresentations that characterize the materialabsorption. The absorbing impedance sheets usedin the three layer model require a layer of freespace on either side; consequently, the one layermodel requires a different material representationthat can model readily be combined with themetallic outer layer. The difficulty in representingthe entire vehicle in one layer is the contrastbetween the aluminum properties and that of theacoustic blankets. Accordingly, an option wasused to apply the blanket properties as a coating tothe metal outer layer. Material properties of thelossy metals and dielectrics are represented in theFEKO material tree. Dielectrics are then selectedin a thin dielectric sheet (TDS) with specifiedthickness. Coatings are selected from the TDSsingle layer options. The TDS is implemented

IV. EQUIVALENT ONE LAYER MODELIt is desirable to further reduce the requiredcomputational resource and run-time requirementsof the three layer simulation by using anequivalent one layer model. Another reason toform a one layer equivalent model is the limitedavailability of vehicle CAD models with blanketconfiguration information.

The aluminum foil outer layer and acousticblanketing layers were represented within FEKOas desclibed below:

• The fairing outer walls were representedas a single layer lossy metal with athickness representing the industryaluminum foil that lined the prototypefairing (0.127 mm thick).

• Kapton sheets are modeled with a surfaceimpedance based on industry data at themodel frequency.

• Gaps between the impedance sheetsrepresent the foam layer.

• Free space is required on both sides of theimpedance sheet thus a thin layer of freespace is implemented between the Kaptonand aluminum.

Test to computational model comparisonspresented in Fig. 4 show favorable results over thefrequency range considered. The averagevariation was 2.43 dB from test data. This isreasonable for a test article to model comparisonsgiven uncaptured variations in test set-up.Selection of this frequency range is related to thesupporting data availability for future workcomparisons.

Fig. 3. FEKO model with acoustic blankets.

within FEKO in a similar way as the impedancesheet in (1) with the Zs term desclibed in (2) [6].

A TDS is required to be geomenically orelectrically thin (approximately 1/10 the smallestelement or wavelength respectively). Due to thegeometric constraint, a dominant limitation isoften encountered as the automatic mesh routinegenerates fine elements to accurately characterizerespective geometries. However, if the coating isgeometrically small with respect to the majority ofthe elements, the geometric constraint iseffectively ignored in the solution. A FEKOutility will perform a validate check, and onlywarnings will be returned in the solution. It is alsoimportant to note that the electrically thinconstraint is relative to a wavelength in theinterfacing medium, but the layer does not have tobe electrically small relative to a wavelength of thelayer itself [7]. Nevertheless, it is often thesituation that the actual thickness of the blanketscannot be represented in the coating, and anequivalent method must be demonstrated andevaluated.

B. Sample s-parameter measurementThe one layer coating constraint drives the need

to represent the three layer blanket model in thewaveguide with a one layer TDS. The NicholsonRoss Weir (NRW) technique is used to derive anequivalent permittivity of the entire layeredblanket using s-parameter measurements. Ablanket sample was placed in an S-Bandwaveguide. The s-parameters are then measuredwith a vector network analyzer as in Fig. 5. Theseparameters are then evaluated in expression (3) toprovide an approximation of an equivalentpermittivity of a homogenous sample of the samelength. As most launch vehicle blanketingmaterials are non-magnetic, setting thepermeability, Ilr, to one simplifies the pelmittivitydetermination. Moreover, TDS implementationrequires the pem1eability to be continuous with thesurrounding media.

E = A~ (2. _[_1 In (~)]2) (3)T Jlr A~ 21fL T

Where, leo is the freespace wavelength for thedesired frequency, Icc, is the waveguide cut-offwavelength, L is the sample length, and T is thetransmission coefficient determined by themeasured S-parameters [11].

Fig. 5 Material sample test fixture.

Detennining the permittivity of a homogeneoussample using waveguide measurements andcomputational models has been verified aseffective in the literature [12]. In this paper, theNRW technique is used to determine a first levelapproximation of an equivalent pennittivity thatwould apply to a dielectric block with the samemeasured s-parameters, although the sample itselfis layered. Full wave analysis is then used tomodify the permittivity at each frequency until asufficiently close approximation of the s­parameters is found. This equivalent permittivitydata is then used to construct the coating in thevehicle one layer model.

C. Waveguide sample modelsA three layer MoM model was first

constructed in FEKO as shown in Fig. 6 toemulate the actual S parameter measurement set­up. The permittivity and conductivity of eachKapton layer was characterized as a dielectric withthe thickness accounted for in the TDSimplementation. The foam was represented by airas in the three layer vehicle model.

Fig 6. FEKO MoM model of vehicle blanketsample.

Model 521 Comparison

Fig. 9. Waveguide S-parameter test datacompared to FEKO models

D. Equivalent one-layer vehicle modelResults in Fig 10 shows that incorporation of

the pennittivity and loss tangent derived from theNRW waveguide technique into a TDS coating ofa single metal layer in the vehicle model providesa reasonable con-elation to the test data, as doesthe 3 layer model. First, the original samplethickness results are applied directly to the coatingproperties. Due to layer wavelength relatedconstraints, however, the thickness of the coatingis set at 3 skin depths of the kapton layer. A closerapproximation is achieved by using (3) to providea different permittivity and loss tangent tocon-espond to a sample thickness adjusted to asmaller value. Results shown are for a TDS lengthof 1/6 of the original sample which varied fromthe test results an average of only 2.5 dB.

~ o.

.~

a03~

Fig 7. Equivalent homogeneous dielecttic block.

It is straight forward to convert the separatelayer model into a multilayer TDS which only usesone face in the geometry representation.However, the multilayer TDS cannot berepresented as a coating to a metal. Hence,representation of the material in a single TDS ispursued.

FEM was employed to verify that the NRWderived equivalent properties derived with (3)represent the S parameters when the waveguide isfilled with a homogeneous dielectric block. TheFEM model in Fig. 7 effectively reproduced theresults as shown in Fig. 9 with some parameteroptimization in the model. In this instance theregions defining the boundary of the block arerepresented as the dielecttic material andimplemented with permittivity parameters withrespective loss tangents.

The parameters were implemented with a TDSsingle layer as shown in Fig. 8 for finalimplementation into the vehicle fixture.

------,---Fig. 10 Single and three layer vehicle model

comparisons with test data.

The upper and lower bounds represented in fig10 are based on cavity Q equations for aluminumand blanketed walls [13]. It is evident that the

·····.... 1'

lJOll1l1!Il10l

Equivalent layer Comparison

~:...;~- _..~- ----- -~:--::

...... _,,-• ~..._.__._.......... •J.

01

001

1'l:IJ!

,! 0,«11

I 1£·?i

IE"

1[<17

\['

Ill)

Fig 8. TDS layer in waveguide

When meshing constraints require a reducedthickness in the TDS layer, a thinner layer can beestablished by changing the sample length in (3) toachieve a cOlTesponding pelmittivity. Fig 9.shows a comparison of test, MoM separate layermodel, FEM dielectric block model and the finalsingle layer TDS with original and reduced samplethicknesses. Variation of the material parameterscan then be exercised to provide a closer match tothe original S21 measurements.

FEKO models provide a significant improvementover relying on approximation results. It shouldbe noted that the primary intent of the Q relatedapproximations are to evaluate chambers with veryconductive walls with small absorbers present, butthe application of these equations are oftenextended cavities with more complex materialconfigurations.

The efficiency benefits of using MLFMM in athree and one layer model as compared to MoMare shown in Table I.

Table 1: Memory/Run Time ComparisonFreq " unknowns ICPU ICPU Time Peak Me(GHz) Time/pro All All

cess (hrs) processes I Processes(hrs) I(GB)

MOM 1 laver 2.6 124.377 21.2 339 115

MlFMM 3 2.6 372,622 3.9 60.9 10laver

MLFMM 1 2.6 124,377 .066 1.1 2.2lave'

V. CONCLUSIONThis paper shows that fairing structures with

complex blanketing materials can be modeledeffectively with equivalent impedance techniquesin a multilayer MLFMM model within the FEKOsolution environment. This is important becausequantifying fields due to transmission within avehicle fairing has largely relied on generalreverberation chamber average powerapproximation. The techniques explored herewere the three layer and one layer models. Fromthis data set, both methods appeared to have animprovement over the power approximationtechniques for a launch vehicle with simulatedacoustic blankets. The equivalent one-layerapproach utilized a novel application of NRWformulations to derive an equivalent permittivityof the three layer configuration. Future workincludes extending the frequency range beyond S­Band and the application of this technique to otherlayered materials such as composite vehiclestructures.

ACKNOWLEDGMENTThe testing of the blanketing layers was

performed by Dr. Ellen Lackey and Dr. ElliottHutchcraft from University of Mississippi.

REFERENCES[1] R. Brewer and D. Trout, "Modern Spacecraft,

Antique Specifications," 2006 IEEEInternational Symposium on ElectromagneticCompatibility, pp 213-218, Aug, 2006.

[2] D. H. Trout, J . E. Stanley, and P F. Wahid,"Electromagnetic Launch Vehicle Fairing andAcoustic Blanket Model of Received Powerusing FEKO." ACES Conference, March2011.

[3] D. H. Trout, P. F. Wahid, and J. E. Stanley.,"Electromagnetic cavity effects fromtransmitters inside a launch vehicle fairing,"IEEE EMC Symposium Austin: IEEE, 2009,in Proc of IEEE EMC Symp. on EMC,Austin, 2009, pp. 70-74.

[4] M. Kandula, K. Hammad, and P. Schallhorn,"CFD Validation with LDV Test Data forPayloadlFairing Internal Flow," Proc. AlAA,2005-4910-9151.

[5] ETS Lindgren Website on double ridgedwaveguide horn, EMCO 3115, User Manual.[Online]. Available: http://www.ets­lindgren.com/manuals/3115.pdf.

[6] FEKO User's Manual, July 2008.[7] Jakobus, u., "Comparison of different

techniques for the treatment of lossydielectric/magnetic bodies within the methodof moments formulation," AE"VInternational Journal of Electronics andCommunications, 2000, Issue 3, Vol. 54, pp.163-173.

[8] U. Jakobus, J. Tonder, and M. Schoeman.,"Advanced EMC modeling by means of aparallel MLFMM and coupling networktheory," IEEE 2008.978-4244-1699-8/08.

[9] J. E. Stanley, D. H. Trout, S. K. Earles, 1. N.Kostanic, and P F. Wahid., "Analysis ofMulti-Layer Composite Cavity UsingFEKO." ACES JOURNAL, January 2010,Issue 1, Vol. 25, pp. 69-74.

[10] W. C. Chew, J. M. Jin, E. Michielssen, and J.Song, Fast and Efficient Algorithms inComputational Electromagnetics, ArtechHouse, Inc., Norwood, MA, 2001.

[11] Rhode & Schwarz, Measurement ofdielectricmaterial properties. s.l. : Application CenterAsialPacific, 2006. RAC0607-0019.

[12] M.D. Deshpande, C.J. Reddy, Application ofFEM to estimate complex permittivity of

dielectric material at microwave frequencyusing waveguide measurements, NASA,Langley Research Center, 1995.

[13] D. A. Hill, M. T. Ma, A. R. Ondrejka, B. F.Riddle, M. 1. Crawford, and R. T. Jonk,"Aperture excitation of electrically large,lossy cavities," IEEE Trans. Electramagn.Campat., vol. 36, no. 3, pp. 169-177, Aug.1994.

Dawn H. Trout (M'95) hasbeen a member of IEEE since1995. She received her B.S.degree in Electrical Engineeringfrom Memphis State Universityin 1989 and her Masters inElectrical Engineering from theUniversity of Alabama in

Huntsville in 1995. She is currently pursuing herPhD in Electrical Engineering at University ofCentral Florida through the Kennedy Space CenterGraduate Fellowship Program.

She has worked at NASA in the area ofelectromagnetic compatibility for twenty years.She has served as lead of electromagnetic teams atMarshall Space Flight Center in Alabama and atKennedy Space Center in FIOl;da. She hasinitiated multiple electromagnetic related studiesin her career and her current research interestsinclude electromagnetic fields in large compositecavities and indirect lightning effects.

Ms. Trout has served on multiple EMCstandards committees and led the development ofISO 14302, Electromagnetic Compatibility, SpaceSystems Standard for which she received an AIAAaward.

Dr. James E. Stanley (M'87)has been a member of IEEEsince 1987. He received hisB.E.E. from Auburn Universityin 1988 and his Masters ofScience in Engineering(Electrical Engineering) fromMercer University in 1997. Hereceived his Ph.D. in Electrical

Engineering from the Florida Institute ofTechnology in 2010.

He has worked in DoD on multiple airframesand subsystems. After moving to the commercial

sector with General Electric and communicationcompanies, he is contributing to the Qinetiq-NorthAmerica team as a NASA contractor. For the lastfive years he has been the lead in electromagneticanalysis for the Kennedy Space Center LaunchServices Program. He has received numerouscommendations for bringing the ElectromagneticCompatibility Team from a purely back-of-the­envelope approach to one with intensive numericalanalytical analysis capability. His skills areincreasingly in demand to evaluate missionintegration issues with antennas in cavities. Hehas done extensive research on layered materialsand modeling these layers with impedance sheets.He also has expertise in developing computersystems and writing electromagnetic modelingcodes.

Dr. PARVEEN F. WAHIDreceived her B.S. degree inMathematics and Physics in1969, her M.S. degree inPhysics from the University ofMysore, India in 1971 and herPh.D. in ElectricalCommunication Engineering

from the Indian Institute of Science, India, in1979.She was a Research Associate at the ElectricalEngineel;ng Department, University of Utah from1980-1982 and at the Electrical EngineeringDepartment, University of Nebraska, Lincoln from1982-1983. Since 1984 she has been with theUniversity of Central Florida, where she is now aProfessor in the department of ElectricalEngineering. She teaches electromagnetics,antenna theory and design and microwaveengineering courses. Her research interests are inthe area of the design of microstrip antennas andarrays and adaptive arrays for wirelessapplications and she has over 50 technicalpublications.

In 1989 Dr. Wahid was named the Tau Beta PiProfessor of the Year. She received the College ofEngineering Excellence in Teaching Award in1994 and 1999. In 1991 she received theUniversity of Central Florida Excellence inAdvising Award and in 1997 the University ofCentral Florida Excellence in Professional ServiceAward In 2000 she was awarded the IEEE Region3 Outstanding Engineer Educator Award and the

IEEE Florida Council Outstanding EducatorAward. She is a recipient of the IEEE MillenniumAward. She was the Technical Program Chair forthe 1999 IEEE International AP/URSI symposiumand the General Chair for the 1998 IEEE Region3 Southeastcon conference. She has served manytimes on the technical program committee for theIEEE AP/URSI conferences. Dr. Wahid is aSenior Member of the IEEE and a member of theEta Kappa Nu and the Tau Beta Pi Societies.