advanced materials technology · pdf fileadvanced materials technology insertion ......

L. W. HUNTER ET AL.

APPLIED RESEARCH

Advanced Materials Technology Insertion

Lawrence W. Hunter, Donald D. Duncan, Frank F. Mark, James S. O’Connor,and James W. White

his article highlights the accomplishments of an independent research anddevelopment effort undertaken in response to the increasing importance of advancedmaterials to the Laboratory’s future work.

T

INTRODUCTIONThe use of advanced structural materials in missiles,

satellites, ships, and submarines—all systems of impor-tance to the Laboratory—is rapidly increasing. Ad-vanced materials offer lighter weight, higher strength,and improved resistance to heat and chemical attack;some lend themselves as agents for certain active con-trol functions (by dimensional or viscosity change).The payoff is improved performance of systems appli-cations, including higher thrust-to-weight ratios foraircraft and rockets and improved resistance of satellitesto laser and nuclear radiation.

The Advanced Materials Technology Insertionproject was launched to advance APL’s capabilities tomodel, design, fabricate, and test prototype advancedmaterials/structures to enhance performance through-out their life cycle. The project was active as aLaboratory-wide thrust area and was supported by in-dependent research and development funds from Jan-uary 1990 through September 1994. To achieve diver-sification, the project (coordinated by L. W. Hunter)comprised a series of small, mainly independent taskswhose specific objectives were modest. Each one, how-ever, made a unique contribution. All eight tasks fo-cused on structural materials rather than optical andelectronic materials, and addressed the interactions be-tween the materials and their environment. Completedtasks and contributors are listed in the boxed insert.

358 JOH

Task 1 demonstrated a special technique needed tofabricate vessels for use in deep sea environments,where the pressure tends to collapse the vessel. The tasksuccessfully showed approaches for fabricating the es-pecially difficult geometries of intersecting cylindersand fittings.

Task 2 focused on the fabrication of copper–graphiteradiation shields to protect spacecraft electronic com-ponents from nuclear and ionizing radiation. Theshields have favorable structural characteristics, meetrigorous outgassing requirements for space qualifica-tion, are lighter than conventional copper metallicshields, and can be customized for an anticipated radi-ation environment.

In parallel, Task 3 enhanced our in-house capabilityin heat pipe applications. Heat pipes carry heat rapidlyfrom a hot to a cool region without the need formechanical pumping of the coolant. Successful thermalmanagement results in survivability of materials, in-creased strength, reduced corrosion, and reduced ther-mal stresses (as structures are isothermalized).

In Task 4, a trainable, quick diagnostic system wasdeveloped to advise operators performing nondestruc-tive inspections of aircraft, new ship structures, andother critical composite structures. Neural networkalgorithms were taught to recognize individual, inter-laminar flaw structures from ultrasonic scans. This

NS HOPKINS APL TECHNICAL DIGEST, VOLUME 16, NUMBER 4 (1995)

activity progressed to on-line diagnostics and controlwork in which the patterns of real-time sensor signa-tures were analyzed with neural network feature extrac-tors and expert systems in order to perform processcontrol and to support condition-based (as opposed tocalendar-based) maintenance.

Highlights of the remaining tasks are described inmore detail in the following sections. Readers interest-ed in other recent Laboratory accomplishments inmaterials research and applications (covering structur-al, optical, and electronic materials) are referred tovolumes 13(3) and 14(1) of the Johns Hopkins APLTechnical Digest.

NEW METHODS FOR TESTINGMATERIALS EXPOSED TO HEATAND OXIDATION

This task focused on the fabrication of a novel ap-paratus to evaluate the performance of materials atultrahigh temperatures in the presence of oxygen. Newmaterials that can resist heat and oxidation are neededin future high-speed vehicles. The materials consideredhere resist heat and oxidation by forming an adherent

COMPLETED TASKS IN THE ADVANCEDMATERIALS TECHNOLOGY INSERTION PROJECT

Task titles are followed by the names of the principal inves-tigators (PIs) and main contributors (in alphabetic order; af-filiations are also noted when other than APL).

1. Composite Pressure Vessels—Intersecting Cylinders withFittings: P. J. Biermann (PI), S. Cooper, and G. Dailey

2. Reduced-Weight Satellite Radiation Shields from Compos-ites: J. J. Suter (PI), J. C. Poret, and M. Rosen (JHU, Dept.of Materials Science and Engineering)

3. Extending Materials Performance in Hypersonic Platforms:F. G. Arcella (PI), S. Corda, M. A. Friedman, L. W. Hunter,and J. B. Kouroupis

4. New Advisory Tools for On-Line/In-Field Inspection ofComposites: F. G. Arcella (PI), P. J. Biermann, L. Brown(David Taylor Research Center), M. A. Friedman, P. C.Lebowitz, and R. W. Newman

5. New Methods for Testing Materials Exposed to Heat andOxidation: L. W. Hunter (PI), D. A. Carpenter, M. A.Friedman, J. R. Kuttler, F. F. Mark, R. M. Sova, andM. E. Thomas

6. Noncontact Strain Measurements: D. D. Duncan (PI),L. W. Hunter, S. J. Kirkpatrick, and F. F. Mark

7. Novel Smart Materials and Structures Made with Electro-rheological Fluids: L. W. Hunter (PI, FY91–93), F. F. Mark(PI, FY94), F. G. Arcella, P. J. Biermann, J. Coulter (LehighUniversity), M. D. Donohue (JHU, Dept. of Chemical En-gineering), M. R. Feinstein, J. R. Kime, D. A. Kitchin,D. R. Kuespert (JHU, Dept. of Chemcial Engineering),J. S. O’Connor, and B. Platte

8. Nonmetallic Motor Liner Materials: J. W. White (PI) andP. J. Biermann

JOHNS HOPKINS APL TECHNICAL DIGEST, VOLUME 16, NUMBER 4 (19

ADVANCED MATERIALS TECHNOLOGY INSERTION

solid-oxide coating (passive oxidation). The rate offormation and the adherence properties of the oxidedetermine the lifetime and maximum speed of thevehicle.

The original motivation for this task was the needto develop materials for the combustor liner of anew ramjet concept being designed by the Laboratoryfor use onboard ships.1 Ramjets carry fuel but no ox-idizer, and must be boosted to a speed at which the airintake is adequate for net thrust. However, ramjetshave two advantages over rockets for shipboard use:(1) a greater powered range and (2) the option tothrottle the engine and retarget the missile in flight.For a given launch tube volume, a ramjet flies a flattrajectory, which takes it farther and faster than arocket, whose ballistic trajectory can be 250 mi highfor a 600-mi range.

Predicted combustion temperatures for plannedmissions exceed 3000 K, so carbon becomes the can-didate material for the combustor walls (the strengthof carbon materials, unlike metals, can increase astemperature rises). However, engine cycle analysespredict that optimum thrust will occur with excess air,creating a combustor environment in which carbonburns. A few materials (namely ceramic oxides) canresist oxidation at 3000 K, but they unfortunately reactwith carbon and cannot form a bond.

Several organizations participated in the search forramjet combustor liner materials,2 mainly under spon-sorship of the White Oak Naval Surface WeaponsCenter (so called at the time). In addition, the Lab-oratory’s Milton S. Eisenhower Research Center solvedseveral basic research problems.3 The most promisingidea was to coat the carbon with a ceramic carbide.4,5

Arc heater tests6 showed that hafnium carbide, inparticular, formed a protective oxide scale at 3000 Kon the side away from the carbon. In addition, a rea-sonable carbide thickness could be expected to protectthe underlying carbon for one 10-min ramjet flightbefore the carbide was completely oxidized and attackon the carbon began.

The laser heating apparatus fabricated in this task(Fig. 1) enables similar candidate materials to bescreened and evaluated at temperatures up to 3000 K.Specifically, it allows oxidation measurements to bemade with better control of experimental variablesthan more expensive arc heaters, vitiated air heaters,and other laser facilities. The heat source is theLaboratory’s CO2 laser, which can be varied up to570 W continuous wave. The apparatus uses a self-aligned optical bench containing beam-integrationoptics and a pyrometer tied to an automatic data ac-quisition system. The delivered heat flux is uniformover a 9 3 9 mm square, large enough to eliminatelateral heat conduction effects at the center. Pyrometermeasurements agree with readings from imbedded

95) 359

L. W. HUNTER ET AL.

Figure 1. APL’s laser heating capability. The delivered heat flux reaches 700 W/cm2 = 4.3 BTU/in2·s = 625 BTU/ft2·s; blackbodytemperature reaches 3260 K (≈5440°F). The apparatus has the following advantages: uniform heat flux over a 9 3 9 mm area; accuratetemperature control; long exposure times; low background thermal reradiation, which permits speckle strain measurement at highertemperatures; and noncontact heating.

C o n v e x m i r r o r

I n t e n s i t y p r o f i l e ( a f t e r )

I n t e n s i t y p r o f i l e ( b e f o r e )

B u r n p a t t e r n

2 4 6

T i m e ( m s )

T o p h a t p r o f i l e t o w i t h i n 1 0 %

5 7 0 - W c o n t i n u o u s - w a v e

l a s e r

thermocouples. The atmosphere control box (Fig. 2)also accommodates the collection of laser speckle forsimultaneous thermal expansion or creep measure-ments by a new technique described in what follows.

As a demonstration, hafnium carbide was oxidizedin ambient air in the apparatus. The specimen wasidentical to those exposed for 10 min in earlier archeater tests,6 but here the focus was on its initialresponse.7,8 Interestingly, a new phase was detected,and the total depth of oxidation exceeded predictionsbased on a model that agreed with the longer arc heatertests. The chemical composition of the new phaseand the discrepancy in the depth of oxidationremain unresolved.

Related work on the theory addressed temperaturegradients with variable thermal conductivity9–12 andthe development of a novel and more efficient algo-rithm for calculating the internal temperature distribu-tion of an ablating material.13 The algorithm appliesto materials that gasify at a known temperature ormelt when the liquid is immediately removed. Thechallenge here is that the location of the exposedsurface is unknown in advance. The approach devel-oped makes possible a single integration grid, which

360 JOH

does not need to adapt to the changing surfacelocation. All the nonlinearities of the problem are com-bined into a single functional relationship between theenthalpy and a generalized temperature.

NONCONTACT STRAINMEASUREMENTS

The adherence of high-temperature protective coat-ings is often tied to their thermal expansion, strain, andcreep behavior; a good match with the substrate isneeded to avoid crack formation and disbonding, espe-cially under thermal cycling. The same properties areimportant for fibers and (single-crystal) whiskers usedto strengthen composite materials. Hence, the expan-sion, strain, and creep properties of candidate advancedmaterials must be screened. The difficulty in the ramjetapplication discussed previously is that the data areneeded at temperatures as high as 3000 K, so any clampsmust be well outside the hot zone.

An older approach was to scratch two marks on thesurface of the test material and track the changingdiffraction pattern produced as the material’s dimen-sions changed. One disadvantage of such an approach


Figure 2. This high-temperature controlled atmosphere chamberprovides control of composition and pressure and accommodateslaser heating; noncontact measurement of creep, strain, andthermal expansion; and surface temperature pyrometry. This stain-less steel chamber measures 6 3 7 3 15 in.

Electricalfeedthroughs

Bleedport

Controlledflow inlet

Tubefittings

Tension shaftand bellows seal

High-temperaturelaser beam axis

Speckleinterferometer

laser axis

Zinc selenidewindow

Sapphirewindow

Pyrometerviewing port

Pumpoutlet

Speckle interferometercharge-coupled devicecamera viewing port

is that the scratches can initiate cracks. Another is thatthe measurements apply at only one scale, namely, thedistance between the scratches. The approach we prefertracks the motion of the “speckle” interference patternproduced when an incident laser beam interferes withthe microscopic surface roughness comparable in size tothe wavelength of the light. The pattern would beunderstandably difficult to model, but its motion isreadily measured and is simply related to the material’ssurface motion, which is of interest. We proved thatone speckle measurement is equivalent to many mea-surements using pairs of scratches at different spacings;thus, the statistics of our measurement are greatly im-proved. In addition, we developed a novel data process-ing method that is insensitive to vibrations and turbu-lence, fast, and more accurate. Other advantages of ourtechnique are its compact design and modest resolution



requirements. The speckle pattern is observed with alinear array detector.

Our most recent work is on the measurement ofstrain rates in fibers typically used to reinforce compos-ite materials. The strains of interest are on the orderof tens of parts per million and less. Because the req-uisite forces on these fibers are small (fibers are typicallyon the order of 10 to 100 mm in diameter), we designedand constructed our own microtensile test machine(Fig. 3). It is made of aluminum and stainless steel, witha micrometer head at one end and a piezoelectric trans-ducer (PZT) at the other. The fiber is clamped betweentwo collet-type chucks, one in contact with the mi-crometer head and the other rigidly attached to theload cell. The load cell in turn is attached to the PZT.The load cell is a tension/compression device with amaximum load range of 1 kg. Its output is first inputto a full bridge network and then to a PC. The PZTis equipped with a displacement sensor, allowing it tobe run in a closed-loop configuration with a maximumexcursion of 40 mm, and maximum compressive andtensile forces of 100 and 10 N, respectively. In an actualexperiment, a fiber is secured in the collets, and themicrometer head is backed off to allow preloading ofthe specimen. A sawtoothed voltage waveform is thenapplied to the PZT, causing a periodic strain variationin the fiber.

Our measurement concept is as follows: The objectunder tensile stress is illuminated sequentially with acollimated laser beam at off-axis angles of ±us andobserved normally with a linear charge-coupled devicearray camera; the illuminated portion of the specimenconstitutes the gauge size.

Yamaguchi14 has shown that the speckle motion ob-served at angle u0 for illumination at angle us is givenby

d u uuu

u

u uu

u

euu

uuu

x aLL

aL

L

L

x

z

xx y

( , )coscos

cos

cos sincos

sin

sincos

tancoscos

,

00

2

00

0

00

00

00

1

ss

s

s s

s

s s

= +

− +

− +

− +

V

(1)

where

ax = an in-plane motion,az = an out-of-plane motion,exx = the linear strain in the plane of the laser beam

and detector,Vy = a rotation about the axis perpendicular to the

measurement plane,

95) 361

L. W. HUNTER ET AL.

Figure 3. Microtensile test machine.

L0 = the observation distance, andLs = the source distance.

By observing normally (u0 = 0), using collimated laserbeams (Ls ‰ `) and subtracting the speckle move-ments observed for two equal but opposite source an-gles, we arrive at the formula for differential specklemotion (before and after stress):

d d u d u e uA x x Lx xx≡ − − = −( , ) ( , ) sin .0 0 2 0s s s (2)

This result suggests that, by measuring the differentialspeckle motion, one can infer the strain, exx.

Figure 4 is a schematic of the measurement config-uration. The entire experiment is managed by an IBMPC. The camera is controlled by means of a plug-in cardon the computer bus. Integration time, sample rate, anddata acquisition are all adjustable in software. Straindata from the load cell are recorded using a plug-inanalog–digital card. This same card produces (undersoftware control) the square-wave voltages used to trig-ger the shutters and the camera.

As suggested previously, the crux of the problem isto determine the speckle pattern shifts resulting fromthe strain. Traditionally,15,16 these shifts have beendetermined by calculating the cross correlation be-tween the two patterns and pinpointing the location ofthe peak value. We have taken an entirely differentapproach that uses what we call a speckle history.17,18

Figure 5 shows a typical speckle pattern as recordedby a linear array camera. When a sequence of specklepatterns such as this is combined in an array, one ob-tains a spatiotemporal history of the speckle patterns.A gray-level display of such a speckle history, shown inFig. 6, is a sequential record of one-dimensional speckle

362 JOH

patterns such that the spatial di-mension is horizontal and the tem-poral axis is vertical. For this mea-surement, the object underwent aslow linear strain from 210 to +10me. Here, e signifies “strain,” so thatme denotes microstrain, or parts permillion. In this display, one per-ceives the slow movement of thespeckle pattern.

The information we seek is thetime rate of speckle pattern move-ment, which is reflected in the tiltof the corrugated structure. A con-venient means of extracting an es-timate of this tilt is to perform atwo-dimensional fast Fourier trans-form (FFT). Figure 7 shows theresult of such an operation. In thisfrequency domain, the horizontal

axis is spatial frequency and the vertical axis is temporalfrequency. Zero frequency (“DC”) is in the center. Thebright line passing through zero frequency is perpendic-ular to the corrugated structure shown in Fig. 6. Itsslope, therefore, is simply the time rate of specklepattern shift. By taking the difference of these slopesas seen for the two illumination angles, the (time rate

–us

M1

Beamsplitter

Laser

Fiber

Strain Strain

+us

M2

S1 S2

Charge-coupleddevice

Figure 4. Measurement configuration. S1 and S2 are shutters, M1and M2 are mirrors, and us denotes the off-axis angle.


Pixel number

Gra

y le

vel

0 64 128 192 256 320 384 448 5120

64

128

192

256

Figure 6. Display of speckle history.

Figure 5. Typical speckle pattern.

of) strain can be inferred through Eq. 2. Specifically, weuse the time derivative of that equation:

d d u d u e u˙ ˙( , ) ˙( , ) ˙ sin .A x x Lx xx≡ − = −0 0 2 0s s s (3)

Inversion of this equation yields the estimate for thestrain rate:

˙ ( )sin

,euxx

m mL

= −2

s

1

02 (4)

where m1 and m2 represent the slopes as calculated forthe two illumination angles.

In principle, the spectral transformations can be per-formed using an FFT. However, we have had betterresults using an FFT in the spatial dimension and aparametric estimator19 in the temporal dimension.



Specifically, we use an autoregressive estimator (mod-ified covariance method) on 10 to 20 records with3 to 5 poles. One characteristic of such parametricestimators is their ability to incorporate a priori knowl-edge about the form of the computed spectrum. Inour case, we know that the focused speckle historywill consist of a well-localized bright line runningthrough the DC point. In addition, parametric modelssuch as these tend to produce very “spiky” spectra,thus allowing good resolution with short data records.This characteristic, in turn, helps us analyze signals thatare not stationary, such as those observed when thespeckle patterns, for sufficiently large strains, becomedecorrelated.

This approach to evaluating speckle translation is astraightforward generalization of the method ofSharpe20 (see the boxed insert, Laser Speckle). In thistechnique, a hardness tester is used to produce a pairof indentations in the material. Illumination by a co-herent source then produces Young’s fringes (see boxedinsert on same), which are monitored with a pair ofdetectors. As in our approach, the differential config-uration allows the subtraction of rigid body motions.

Our speckle history approach to processing Sharpe’sdata (from each detector) would produce a single pairof bright spots, symmetrically arrayed about zero fre-quency, in the frequency plane. The slope of the lineconnecting these two bright spots is the time rate ofYoung’s fringe shift.

Alternatively, the two delta function sources (theilluminated indents from the hardness tester) yield acosine field distribution in the far field, i.e., a Fourierrelationship exists between source and field distribu-tions.21 A second (two-dimensional) Fourier transform

S p a t i a l f r e q u e n c y

T e

m

p

o

r a

l

f

r

e

q

u

e

n

c

y

Figure 7. “Focused” speckle history.

995) 363

L. W. HUNTER ET AL.

LASER SPECKLELaser speckle is an interference phenomenon observed

when a coherent source of electromagnetic radiation is reflect-ed from a surface that is rough compared with the wavelength.It creates a speckled or wormy appearance. The term is some-what misleading because the phenomenon is observed in avariety of situations such as synthetic aperture radar imagery.Shown in the figure (top) is a way of observing so-called “ob-jective” or nonimaged speckle. A rough object is illuminatedwith a laser source, and the resulting scattered radiation isobserved with a camera without a lens. Also shown (bottom) isthe resulting pattern as displayed on a TV monitor. The sizesof the speckles are a function of the wavelength and observa-tion geometry, and can be described quantitatively usingYoung’s fringe analogy (see the boxed insert, Young’s Fringes).

Subjective speckle is observed when the illuminated objectis imaged (by the eye, a camera, film, etc.). In this case theimage of the object appears mottled or speckled.

DetectorSample

Laser

illuminatio

n

Laser speckle pattern.

364 JOH

(computed on the measured intensity) inverts thisrelationship, thus reproducing three delta functions.Because of the detection process, the focused line is notan image of the illumination stripe, but rather its au-tocorrelation.22 For this simple case, therefore, weobserve a pair of bright spots off axis and a third brightspot at DC.

In the objective speckle technique, the complexspeckle pattern can be viewed as arising from all pos-sible scatter pairs on the rough surface. Thus, eachpoint in the focused speckle history corresponds to aunique scatterer separation on the object.

If the object under stress is straining equally at allscale sizes, the bright line in the frequency domain willbe straight; however, without equal straining over theregion of illumination,17 the focused line will displaycurvature. The amount of curvature reflects the scalesize dependence of strain.

The specific parameters used for our measurementsare listed below, and results are summarized in Table 1.The test specimen was a 16-cm-long, 260-mm-dia. ti-tanium wire undergoing a sawtooth variation in strainat a frequency of 7.5 mHz. Amplitudes of the stresswaveforms were varied between two experiments.

Wavelength, 821 nmLaser power, 70 mWSource radius of curvature, ∞Source incidence angles, ±45°Observation distance, 0.52 mIlluminated stripe (gauge length), 2 cmObservation angle, 0°Camera integration time, 350 and 500 msCamera sample interval, 0.5 and 1.0 s

The stress waveform for the first experiment isshown in Fig. 8 and the corresponding speckle historiesin Fig. 9. By following previously developed pro-cedures17,18 using an FFT/autoregressive transformation(20 records, 3 poles), we estimated strain rates of1.995 me/s (±0.018 me/s) in one experiment and0.827 me/s (±0.015 me/s) in the other. Numbers in

Table 1. Experimental results.

Parameter Experiment 1 Experiment 2PZT velocity (mm/s) 0.3804 0.1406Load rate (g/s) 1.144 0.7056Strain rate from load cell (me/s)a 1.863 0.7056Strain rate from PZT (me/s) 2.378 0.8785Strain rate from speckle (me/s)a 1.995, ±0.018 0.827, ±0.015E estimated from speckle (GPa) 106 94.3Differential speckle motion (pixels) 2.7 0.57aAssuming Young’s modulus E = 110.3 GPa.



Geometry for point scatter pair.

s/2

–s/2

R1

R2

x

L0

us

For small changes in separation ds, and the corresponding shiftin the reference position dx, Eq. 4 yields

dd

e uxs

Lx

xx=

= −0

0 sin , (5)

where the strain is defined as the fractional change in thescatterer separation. For a differential viewing geometry, thedifferential Young’s fringe shift is

d d u d u e uA x x Lx xx= − − = −( ) ( ) sin .s s s2 0 (6)

With the procedure illustrated above, the formal results attrib-utable to Yamaguchi14 can be derived heuristically.

YOUNG’S FRINGESTo elucidate the objective speckle technique, we use a

particularly simple geometry, illustrated in the figure. Here weshow a pair of scatterers separated by a distance s, illuminatedwith a plane wave at angle us, and observed normally at adistance L0. Although not necessary, for simplicity we ignoreall rigid body motions. The intensity pattern in the observationplane is given by

I x k R R s( ) cos[ ( sin )] ,∝ + − +2 2 2 1 us (1)

where k is the wavenumber (2p/l). From the figure, we havethe following definitions for the geometric variables:

R L x s

R L x s

2 02 2

1 02 2

2

2

= + +

= + −

( / ) ,

( / ) .(2)

With these definitions, Eq. 1 above becomes in the paraxialapproximation

I xs

Lx L( ) cos ( sin ) .∝ + +

2 2

2

00

pl

us (3)

To quantify the behavior of this fringe pattern as the separationbetween the scatterers changes, we wish to keep track of aposition of constant amplitude,

2

00

pl

us

Lx L C( sin ) .+ =s (4)

parentheses represent 90% confidence intervals. Esti-mates of the strain rates obtained from the load celldata, assuming a Young’s modulus of 110.3 GPa,23 were1.860 and 0.706 me/s, respectively.

By using knowledge of the loading rates, along withour strain rate estimates, we were able to estimate theYoung’s modulus of the titanium fibers at 106.0 and94.3 GPa, respectively. These figures are a bit low forbulk titanium but may be reasonable for fibers of thissize.

How precisely can this approach estimate strain (orcreep) rates? Several factors must be considered. Someare affected by the physical measurement approachand others by the subsequent data processing algo-rithm. Using a detailed error analysis,24 it appearsthat the strain resolution is not imposed by thealgorithm itself, but rather by a number of competingnoise sources, which produce small, very-low-frequencytransients. These include, but are not restricted to, laserwavelength drift; temperature transients in the supportstructure for the camera, laser, and tensile machine;barometric pressure fluctuations; deviations from sym-


40

30

20

10

00 100 200 300 400

1000

950

900

850

Time (s)

PZ

T e

xcur

sion

(mm

)

Load

(g)

Figure 8. Stress waveform for first experiment (PZT = piezoelectrictransducer, black curve; load cell, red curve; shaded area indicatesdata for which strain is estimated).

metry for the two illumination directions; creep in thetest fiber; and slippage of the fiber in collet chucks.

To the extent that these parameters are known orcan be controlled, it appears that the algorithm that wehave developed can accurately measure total strains onthe order of 1 ppm.24

995) 365

L. W. HUNTER ET AL.

Figure 9. Speckle histories for each of the two illumination direc-tions for the first experiment. Data for which strain estimates weremade are boxed.

NOVEL SMART MATERIALSAND STRUCTURES MADE WITHELECTRORHEOLOGICAL FLUIDS

Although the tasks described thus far were closelyrelated, other tasks in the Advanced Materials Tech-nology Insertion project allowed APL to keep its in-vestment diversified and encouraged further collabora-tions across the Laboratory. The task described in thissection concerns “smart” materials and structures,whose properties may be controlled in real time inresponse to their environment. Several applicationswere investigated, including control of optical signa-tures.25 Another application was protection againstshocks and vibrations over a wider range of conditionsthan is now possible. This task resulted in the fabrica-tion of two promising smart structures for shock andvibration protection. The first, a composite plate whose

366 JOH

vibrational modes and frequencies are controllable inreal time, is described in Ref. 26. The second, a shockabsorber with variable set point, is described here.

Our approach was based on electrorheological (ER)fluids, which have become the subject of increasinginterest. An ER fluid is a dispersion of polarizable,dielectric particles in an electrically nonconductingliquid with a different dielectric constant. The rheolog-ical properties of the fluid, especially the yield stress andapparent viscosity, are then controlled by the applica-tion of a strong electric field (either DC or AC), atwhich time the particles line up in strings parallel tothe field. The characteristics of the ER fluid differ ac-cording to (1) compounds selected, (2) particle size,concentration, and density, (3) additives to the basicfluid, (4) stabilizers and enhancers, (5) electrical con-ductivity of the fluid, and (6) electric field strength. EReffects are reversed upon removal of the field. Fluidsprepared for AC applications seem more stable andperform better. Although the mechanism of the ERphenomenon is uncertain, the boxed insert (RheologyModel) provides a simplified discussion of ER behavior.A guide to ER applications is presented in Ref. 27,which compares ER fluids with shape memory alloys,piezoelectric devices, magnetic fluids, and other poten-tial components of new intelligent materials.

To determine the characteristics of ER fluids neededto design the variable shock absorber, a rheometer wasbuilt consisting of two concentric cylinders separatedby a 1-mm gap (Fig. 10). A constant driving force,which can be set over a range of values, moves the innercylinder up with respect to the outer cylinder. Therheometer is filled with the ER fluid, and high DCvoltages are impressed across the cylinders. The forceon the inner cylinder and its displacement are measuredas a function of time at zero field and for constantelectric field intensities. From these data, shear stressas a function of applied voltage and the shear strain rateare determined, and the viscosity of the fluid may becalculated.

Rheometer tests were conducted on a simple ERfluid, namely, a dispersion of diatomaceous earth in aperfluorocarbon fluid base. ER behavior was observedand shear stress measurements made. Their magnitudes,shown in Fig. 11, were small for the intended applica-tion. This and other results led to the following generalconclusions:

1. The ER shear force per unit area (shear stress) gener-ated is small compared with forces normally providedby mechanical devices. To utilize this effect, an ER-based mechanical device should have a large area,operate over a long period of time (e.g., a vibrationdamper), or use a force multiplier device. An exampleof this kind of device is one that uses an ER fluid tocontrol the area of the orifice of a shock absorber.



Figure 10. Rheometer for measuring the shear force exerted by electrorheological fluids.

2. The electric field intensity required to provide asignificant ER effect is about 3 kV/mm. At theseintensities it is difficult to prevent arcing over a smallgap. The electric interference noise produced by in-termittent “micro” arcs somewhat degrades the out-puts of the force and displacement transducers.

3. The most effective fluids are suspensions, but theseare abrasive to moving parts and seals. ER devicesshould be designed to minimize this problem.

The design of the shock absorber reflects the lessonslearned in the rheometer tests. The design concept,

1 . 8

1 . 6

1 . 4

1 . 2

1 . 0

0 . 8

0 . 6

0 . 4

0 . 2

0

– 0 . 2 0 0 . 1 0 . 2 0 . 3

T i m e ( s )

0 V 2 k V

3 k V

D

i

s

p

l

a

c

e

m

e

n

t

(

i

n

.

)

Figure 11. Effect of electric field on rheometer motion.


instead of relying directly on ERfluid effects for shock absorption,applies a conventional hydraulicfluid contained in a thick-walledcylindrical body. This fluid, driventhrough a variable-area orifice by apiston, provides the desired shockcontrol. The area is varied by asliding gate valve coupled to themoving piston shaft through anindependent ER-controlled device.The device initiates the activedamping required when a sensorsystem detects that a given set-level of shock impulse is exceeded.Advantages of this approach are asfollows:

• Only a small amount of expensiveER fluid is required.

• Problems connected with theabrasiveness and degradation ofthe fluid are minimized.

• Few or no design changes are re-quired to switch over to an im-proved fluid.

• Coupling to the piston shaft isstraightforward.

• Accommodation of the high-voltage supply is simple.

• The critical hydraulic fluid properties are well knownand stable.

A prototype shock absorber was fabricated (Fig. 12)and then delivered to John Coulter at Lehigh Univer-sity for performance evaluation, final design changes,and fault correction. A programmable tensile machinewas used for three phases of testing: (1) validation ofthe concept with measurements of the operationalcharacteristics, (2) design and fabrication of the ERcoupling device, and (3) evaluation of the integratedsystem.

For Phase 1, the gate valve was activated manually,the damper was subjected to representative forces andaccelerations, and damper performance was deter-mined. Data obtained were used for the controllerdesign of Phase 2. Initial results showed the maximumdamper forces attained to be of the order of 3000 lbwhen the orifice was closed and about 200 lb whenopen. Figures 13 and 14 give a representative forceversus time plot and peak force versus frequency plot,respectively. The force required for control was foundto be small, only about 10 lb. The concept is consideredproved because the device operated well and did notleak. On the basis of information obtained, an ERclutch was designed and built. Phase 3, the task oftesting the complete system, is now under way.

995) 367

L. W. HUNTER ET AL.

Thus, the shock damper design uses a very smallvolume of ER fluid to minimize abrasion. The workingfluid is an inexpensive, conventional hydraulic fluidwhose characteristics are uniform and well defined.Simple, unique devices can be based on this concept.Validation of the proof-of-principle of the shock damp-er shows that practical applications of this device, bothmilitary and civilian, are possible.

NONMETALLIC ELECTRIC MOTORLINER MATERIALS

This task took the Advanced Materials TechnologyInsertion project in yet another direction, making im-portant contributions toward the development of newmaterials that enhance submersible electric motors.The Navy needs such motors to drive submergedpumps, off-hull shipboard equipment such as trailingantennae and sonar line arrays, and underwater vehiclethrusters. New Navy initiatives use magnetic bearingsin some underwater applications because they can bequieter than mechanical bearings and their controllerscan actively cancel periodic noise sources.

There is also a commercial need to seal motors usedin corrosive or toxic environments. Pumps with at-tached submersible motors can operate while entirelyenclosed within their piping systems, so that no fluidleakage occurs at packing glands or wearing rings.Recently enacted legal restrictions on leakage andenvironmental emissions will increase the demand forthese pumps and the materials that keep them totallysealed.

The key to submersible electric motors is the motorliner; specifically, the inner bore of radial flux submers-ible motors and magnetic bearings with wound statorsmust be sealed to prevent fluid from entering the wind-ings. Metallic liners, often used in submersible motors,are waterproof but impose an eddy current loss. The losscan be a large proportion of the total motor energy. Inone recent example, the eddy current losses in a 50-hpmotor increased the total input power required by morethan 10 hp. The loss is proportional to the thicknessof the metal liner, the strength of the magnetic fieldacross the gap, and the machine’s rotational speed. Itis inversely proportional to the resistivity of the linermaterial. For motor efficiency, especially in inductionmotors, liners must be thin to keep the “air gap” small.

Thus, the goals of this task were to identify appro-priate materials for use as submersible motor liners andto develop processes for applying those materials to theinner bores of test motors. The materials were to be oflow resistivity and thin to minimize the increase in thegap between static and rotating parts. The materialsand processes were to be characterized on the basis oftheir ability to keep moisture out of the motor windingsfor long periods of time. The ideal material would have

368 JOH

Shear stress versus shear strain and shear strain rate for a typicalER fluid.

RHEOLOGY MODELThe figure presents stress–strain relationships typical of

electrorheological (ER) fluids. Shear stress t is plotted againstshear strain g and shear strain rate «g for given electric fieldintensities j. In these plots viscoelastic properties are exhibited.An important parameter is the static yield stress ty,s, which isrequired to initiate flow in the presence of an electric field.Until this stress is exceeded, the material acts as an elastic solidunder shear (the preyield region). It is a characteristic of thefluid regardless of any flow model representation. In the post-yield region (the transition boundary defined by the gcriticalshear strain value), the material becomes a viscous fluid. Stressmagnitudes strongly depend on the electric field intensity. Notethat at zero field, shear stress is low and the fluid is of theNewtonian type.

No model completely describes the observed behavior of allER materials. However, for most cases, the following equationdescribing the Bingham plastic model is sufficiently accurate fordesign work:

t g j t j hg(˙, ) ( ) ˙ .= +

In this equation, the electric field–induced yield stress t(j) isfurther approximated by the dynamic yield stress ty,d, which isthe zero strain rate intercept of the linear fit to the t(«g) data,curve B. The plastic viscosity h is the slope of this fit. Obser-vations show that static yield stresses are higher than dynamicstresses, as curve A indicates. After flow begins, curve B isfollowed as shear strain rate is increased. Then with decreasingstrain rate, stresses fall along curve B and tend toward thedynamic yield stress intercept, characteristic of the fluid de-scribed by the Bingham model. (The interested reader is re-ferred to Ref. 28 for a review of this topic.)

ty, s

ty, d

She

ar s

tres

s (t

)

gcritical

jhigh

Pre

yiel

d

Pos

tyie

ld

Shear strain (g)

jhighShe

ar s

tres

s (t

)

Shear strain rate (g)

Curve A

Curve B

•

j = 0

Slope = h

ty, s

ty, d


Figure 12. Electrorheological (ER) fluid–controlled experimentalshock damper tested at Lehigh University. The piston shaft diam-eter is 2 in. The orifice control rod, shown beside the piston, isactivated by means of an ER device. A 6-in. scale is shown in thefigure.

infinite resistance to moisture penetration and zerothickness. In practical terms, the thickness of the ma-terial would be small relative to the diameter of themotor rotor.

A mylar film sandwiched between two thicknessesof polyester fabric was examined as an initial candidatesealing material. The sample obtained was 0.015 in.thick and would have met the criterion for being thinrelative to the bore diameter of most motors. But

Figure 13. Force versus time at 2 Hz for a variable-orifice shockabsorber.

1000

500

–1000

–500

0

–1500

–2000

–2500

–3000

–35000 0.5 1.0 1.5 2.0 2.5 3.0 3.5 4.0

Time (s)

For

ce (

lb)

Com

pres

sion

Exp

ansi

on

Orifice closed Orifice open



1000

0

–1000

–2000

–3000

–4000

–300

–200

–100

0

100

200

300

1 2

Frequency (Hz)

For

ce (

lb)

For

ce (

lb)

(a)

(b)

Expansion stroke

Compression stroke

Expansion stroke

Compression stroke

Figure 14. Measured performance of the prototype shock damper.Peak forces are plotted versus frequency: (a) orifice closed, (b)orifice open.

attempts to develop the process for attaching this ma-terial to the motor internals showed that the spliceneeded at the cylinder’s longitudinal seam could not bemade without excessive increases in the thickness atthe seam location. An ineffective splice would allowmoisture penetration into the sensitive steel parts of themotor core.

A second material attempted was preimpregnatedglass-reinforced plastic. It was possible to construct acontinuous cylinder of this material without cross-sectional irregularities. The cylinder was constructed byapplying a rectangular sheet of the uncured/curedmaterial to a cylindrical mandrel and wrapping it withheat shrink tape. Heat from the curing cycle shrank thetape, forcing resin to flow into the longitudinal spliceand causing some intertwining of the fibers there.

This initial attempt to manufacture a thin-walledhollow cylinder provided insight into some process dif-ficulties. Resin shrinks as it cures. With the very thinsection of material on the mandrel, the shrinkage couldcause enough tensile stress to fracture the part. In initialattempts, the part failed at its weakest section (thelongitudinal seam), but we subsequently found that theshrinkage problem could be controlled by careful atten-tion to cure rate.

To test the cylinder’s ability to prevent moisturepenetration, it was installed in the inner bore of a small,inexpensive electric fan motor. The motor had anenclosed case and was designed for heat conduction

995) 369

L. W. HUNTER ET AL.

through its exterior. The motor case made it easy toencapsulate the motor winding sections not protectedby the thin cylindrical liner. The encapsulant was apolyurethane similar to the type used in submarinecables.

The motor used inexpensive sleeve bearings. Sinceit was never intended to run under water, it was vul-nerable to corrosion and bearing failure. For the initialunderwater test, the investigators coated the parts withepoxy paint and filled the bearing cavities with water-resistant grease. The small motor was run under waterin a laboratory container for the first test.

The test had two significant results. First, the encap-sulated motor, when run in air, was no hotter than anidentical motor in the original condition. This findinggave investigators confidence in their initial assump-tion that the encapsulant and liner materials would notchange the heat-carrying capabilities of this particularmotor. Second, high potential insulation checks of themotor showed that the thin plastic liner did not preventmoisture penetration. In the original dry condition, aninduction motor’s insulation should provide a resis-tance of at least 2000 MW between winding and case.Most have significantly more. A perfectly sealed sub-mersible motor should have similar characteristics. Agood measure of moisture penetration into the motoris its insulation resistance. After the first 60 h of run-ning submerged, the motor’s measured resistance wasonly 1 MW. This indicated excessive moisture en-croachment, and the test was abandoned.

A third material consisted of a sandwich of two verythin sheets of glass-reinforced plastic with a thin foilof metal between them. The metal foil was 0.003 in.thick, and the total thickness of the sandwich was 0.015in. A new cylindrical shell of the sandwich material wasinstalled on the center bore of an elec-tric motor identical to the one used forthe first test (Fig. 15). After 1969 h ofrunning under water and 289 h underwater deenergized, the motor still hadan insulation resistance of 12,000 MW.The liner material for which the testwas intended appeared to be successful.After its long underwater endurance,this small, inexpensive motor was verycorroded but still functional.

With the success of the liner mate-rial on a small motor, a special appara-tus (Fig. 16) was developed for testingmuch larger (0.5-hp) submersible mo-tors in a continuously cooled watercontainer. The motor selected hadstainless steel corrosion-resistant partsand sealed bearings. Although “splash-proof,” it is not intended for submergedoperation because its windings are not

Figure 16. This stemperature durin

370 JOH

protected at its inner bore. After this motor was treatedwith the new material and processes from the initialefforts, it showed no degradation in extended underwa-ter tests.

Figure 15. This inexpensive motor with badly corroded exteriorremains operable and safe after 2258 h under water.

pecially constructed test cell can change and monitor pressure andg tests.


CONCLUSIONThe Advanced Materials Technology Insertion

project comprised eight different tasks, which en-hanced the Laboratory’s technical and business posi-tion in the field of advanced structural materials. Inaddition, beneficial collaborations were establishedamong staff members in different departments of APLand with the Homewood campus of the University.The project witnessed a growth in materials activitiesat the Laboratory—notably in the Aeronautics Depart-ment, the Technical Services Department, and theResearch Center—and kept alive an awareness of theimportance of advanced structural materials to APL.

REFERENCES1Billig, F. S., “Tactical Missile Design Concepts,” Johns Hopkins APL Tech.Dig. 4(3), 139 (1983).

2Newman, R. W., “Oxidation-Resistant High-Temperature Materials,” JohnsHopkins APL Tech. Dig. 14(1), 24–28 (1993).

3Bargeron, C. B., Benson, R. C., Newman, R. W., Jette, A. N., and Phillips,T. E., “Oxidation Mechanisms of Hafnium Carbide and Hafnium Diboride inthe Temperature Range 1400 to 2100°C, Johns Hopkins APL Tech. Dig.14(1), 29–35 (1993).

4Hunter, L. W., “High-Temperature Chemistry of Materials: An Update,”Johns Hopkins APL Tech. Dig. 11(1–2), 168–174 (1990).

5Hunter, L. W., “High-Temperature Chemistry of Materials,” Johns HopkinsAPL Tech. Dig. 7, 362–371 (1986).

6Hunter, L. W., Kuttler, J. R., Bargeron, C. B., and Benson, R. C., “Oxidationof Refractory Carbides,” in Proc. 10th Annual Conf. on Composites andAdvanced Ceramic Materials, Cocoa Beach, FL (1986).

7Hunter, L. W., and Duncan, D. D., “Improved Control of Oxidation andMechanical Tests at Ultra-High Temperatures,” in Proc. 16th Annual Conf.on Composites, Materials, and Structures, Cocoa Beach, FL (12–15 Jan 1992).

8Wajer, S. D., and Wilson, D. E., Microanalysis of Hafnium Carbide UsingScanning Electron Microscopy (SEM) and Energy Dispersive Spectroscopy (EDS),JHU/APL TEQ/SDW-92-134 (27 Apr 1992).

9Kuttler, J. R., and Friedman, M. A., Oxidation of Hafnium Carbide: ConstantThermal Conductivity, JHU/APL AM-90-P046 (1 Mar 1990).

10Kuttler, J. R., Oxidation of Hafnium Carbide: Variable Thermal Conductivity,JHU/APL AM-90-P061 (3 Apr 1990).

11Kuttler, J. R., Oxidation of Hafnium Carbide: General Surface BoundaryConditions, JHU/APL AM-90-P082 (31 Jul 1990).



12Friedman, M. A., Oxidation of Hafnium Carbide. Calculation of Effect ofTemperature Gradient, JHU/APL AM-90-P140 (3 Jul 1990).

13Hunter, L. W., and Kuttler, J. R., “The Enthalpy Method for Ablation-TypeMoving Boundary Problems,” J. Thermophys. Heat Trans. 5, 240–242 (1991).

14Yamaguchi, I., “Advances in the Laser Speckle Strain Gauge,” Opt. Eng.27(3), 214–218 (1988).

15Barranger, J. P., Two-Dimensional Surface Strain Measurement Based on aVariation of Yamaguchi’s Laser-Speckle Strain Gauge, NASA TechnicalMemorandum 103162, NASA Lewis Research Center, Clevelend, OH(1990).

16Takemori, T., Fujita, K., and Yamaguchi, I., “Resolution Improvement inSpeckle Displacement and Strain Sensor by Correlation Interpolation, inLaser Interferometry IV; Computer-Aided Interferometry, Proc. SPIE 1553, pp.137–148 (1991).

17Duncan, D. D., Kirkpatrick, S. J., Mark, F. F., and Hunter, L. W., “TransformMethod of Processing for Speckle Strain Rate Measurements,” Appl. Opt.33(22), 5177–5186 (1994).

18Duncan, D. D., Mark, F. F., and Hunter, L. W., “A New Speckle Techniquefor Noncontact Measurement of Small Creep Rates,” Opt. Eng. 31(7), 1583–1589 (1992).

19Marple, S. L., Digital Spectral Analysis with Applications, Prentice-Hall,Englewood Cliffs, NJ (1987).

20Sharpe, W. N., “Applications of the Interferometric Strain/DisplacementGauge,” Opt. Eng. 21(3), 483–488 (1982).

21Goodman, J. W., Statistical Optics, John Wiley & Sons, New York (1985).22Goodman, J. W., “Statistical Properties of Laser Speckle Patterns,” in Laser

Speckle and Related Phenomena, J. C. Dainty (ed.), Springer-Verlag, Berlin(1975).

23McClintock, F. A., and Argon, A. S., Mechanical Behavior of Materials,Addison-Wesley, Reading, MA (1966).

24Duncan, D. D., Kirkpatrick, S. J., Mark, F. F., and Hunter, L. W.,“Measurement of Strain Rates in Reinforcement Fibers,” Meas. Sci. Technol.(accepted) (1995).

25Hunter, L. W., Mark, F. F., Kitchin, D. A., Feinstein, M. R., Blum, N. A.,et al., “Optical Effects of Electro-Rheological Fluids,” J. Intell. Mat. Syst.Struct. 4, 415–418 (1993).

26Coulter, J. P., Don, D. L., Yalcintas, M., and Biermann, P. J., “AnExperimental Investigation of Electro-rheological Material Based AdaptivePlates,” in Proc. ASME Winter Annual Meeting, Chicago (6–11 Nov 1994).

27Mark, F. F., Compilation of Adaptive/Smart Material Characteristics, JHU/APLAM-94-S005 (6 Jun 1994).

28Block, H., and Kelly, J. P., “Review Article, Electrorheology,” J. Phys. D:Appl. Phys. 21, 1661–1677 (1988).

ACKNOWLEDGMENT: The Advanced Materials Technology Insertion projectwas supported by independent research and development funds. The charter,objectives, and long-term strategic plans of the project were developed by LawrenceW. Hunter, Richard P. Suess, Frank G. Arcella, Paul J. Biermann, John R.Coleman, and Joseph J. Suter.

THE AUTHORS

LAWRENCE W. HUNTER is a member of APL’s Principal Professional Staff.He holds a B.Sc. (hons.), first class, in chemistry (Carleton University, Ottawa,Canada, 1967) and a Ph.D. in theoretical chemistry (University of Wisconsin,Madison, 1972). He has been employed at APL since 1973 and has worked onmany different interesting projects, typically related to combustion, missilesystems, and the F/A-18. Dr. Hunter coordinated two APL independent researchand development (IR&D) thrust areas and currently helps coordinate Aeronau-tics Department IR&D activities. He has contributed to over 80 refereedpublications including one listed in “The Twenty-Two Most Frequently CitedAPL Publications,” John Hopkins APL Tech. Dig. 7(2), 221–232 (1986). His e-mail address is [email protected].

995) 371

L. W. HUNTER ET AL.

FRANK F. MARK studied physics at the Massachusetts Institute of Technologyand was a member of the Advanced Systems Group of APL’s AeronauticsDepartment. His work focused on fluid dynamics, biomedical research, optics,and advanced materials. Mr. Mark is now retired.

JAMES W. WHITE received a B.S. in engineering physics from the Universityof Oklahoma in 1972. In 1980, the Massachusetts Institute of Technologyawarded him an M.S. in mechanical engineering and the Professional Degree ofOcean Engineer. After completing his career as a naval officer and working inprivate industry for several years, Mr. White joined APL in 1993. He is amember of the Aeronautics Department working in F/A-18 supplier manage-ment. Mr. White is the inventor of an advanced pump, now undergoing Navytesting, that uses a submerged motor drive. His e-mail address [email protected].

DONALD D. DUNCAN is Supervisor of the Measurements and PropagationSection of APL’s Electro-Optical Systems Group. He received his Ph.D. inelectrical engineering from The Ohio State University in 1977. From 1977 to1983, he was employed by Pacific-Sierra Research Corporation, where hemodeled optical propagation phenomena such as aerosol scatter, atmosphericturbulence, and high-energy laser effects (e.g., thermal blooming, aerosolburnoff). Since joining APL in 1983, Dr. Duncan has worked on variousbiomedical engineering projects; provided test program support and data analysisfor a tracking/guidance synthetic aperture radar system; and worked on manymeasurement, modeling, and diagnostic equipment projects in support of thehypersonic interceptor. He also teaches courses in Fourier and statistical opticsat The Johns Hopkins University G.W.C. Whiting School of Engineering. Hise-mail address is [email protected].

JAMES S. O’CONNOR is a member of the APL Principal Professional Staff. Hehas a B.M.E. (mechanical engineering) from the University of Detroit (1966)and an M.Sc. (aerospace engineering) from Ohio State University (1968).Employed at the Laboratory since 1968, Mr. O’Connor has participated in themechanical design of guided missile systems, spacecraft, automated transporta-tion systems, and the Ocean Thermal Energy Conversion power plant. His e-mail address is James.O’[email protected].

372 JOHNS HOPKINS APL TECHNICAL DIGEST, VOLUME 16, NUMBER 4 (1995)

advanced materials technology · pdf fileadvanced materials technology insertion ......

Documents