PERSISTENT SURVEILLANCE FORPIPELINE PROTECTION AND THREAT INTERDICTION

Haibo HuangRich Stephens

Brian VermillionDan Goodin

Bernie Kozioziemski (LLNL)

HAPL Meeting, Livermore, California

June 20-21, 2005

Radiographic Dimension Measurement of Dry DVB Foam

Shells

IFT/P2005-072

Summary

• IFE program uses 4.1 mm O.D. foam shells• Require dimension variations measured to

<1 m• Optical characterization requires immersion

in index-matching fluid.• X-Radiograph system already developed &

tested– Contact images on high resolution film– High precision film digitizer – Analysis algorithm to handle extreme noise

• For the first time, large dry DVB foam shells can be measured

IFE Point Design for DVB Foam Shells

• Sufficient to measure interface radii vs angle to ±1 m

4100 ±200 m

Average Foam Wall: 289 ±20 m Non-concentricity (NC): < 1% *

* NC= (OD/ID Center Offset)/WallAverage

equivalent to WallMax-WallMin < 6 m

** OOR=(ODmax-ODmin)/RadiusAverage

equivalent to Rmax-Rmin = 10 m

CH Wall: (1-5) ±1 m Out-Of-Round (OOR): < 1% **

Foam is difficult to characterize

• Visible light measurements require index matching fluid– Time-consuming– Potential dimension errors

• OD changes by 1-5%• Thicker CH, larger OD change

• X-radiographs are noisy due to large density fluctuations– Obscures interfaces

Radius (um)

Tra

nsm

iss

ion

(a.

u.)

Area AverageSingle line

100 m

wallinterior

X-radiography works when digitized properly

• 12bit, 4 MB, Cooled CCD– Measure whole shell to 0.8

m• Plan APO Microscope lens

– Flat field => CCD compatible– Large N.A. => high resolution

• Type K1a film– Finest grain – Glass substrate => stable

• Software– Noise reduction and rejection– Edge analysis

Accurate interface profiling require lens correction

-3

-2

-1

0

1

2

3

-800 -400 0 400 800Radius from Image Center (um )

Dis

tort

ion

(u

m)

• Must correct lens distortion– Calibrate with stage

micrometer– Verify with circular

standards

• Each lens calibrated separatelyDoes radius and shape

change with position?

3µm pixel error

Foam structure causes unique analysis problems

• Traditional vision-based analysis does not work with low contrast, noisy image– Reduce noise by azimuthal averaging– Reject noise by data correlation– Limit search range

• Interface very wavy with thin overcoat– Flattening (Step 3)

• Extended interface structure– Fresnel simulation determines offset

Capturing edge information

An

gle

Radius

1) Inspect thickness variation by 360˚ unwrapping

2) Sharpen interface with 2nd derivativeD. Bernat, R.B. Stephens, Fusion Technology, V31, P473, 1997

An

gle

Radius

Mapping interface requires careful consideration

3) Sharp outside edge good for auto alignment

4) Reject noise by correlating peak/valley locationsReduces the image to a set of R() files

Separates wavy interfaces => narrows search range

An

gle

Radius

An

gle

Radius

visualize wall thickness variation

Calculating interfaces and walls

Interface Marker (4X)

1200

1300

1400

1500

1600

0 100 200 300

Angle (degree)

Ra

diu

s (

um

)

OD: 3200 ± 3 m

OOR: 0.88 ± 0.04 % DVB Foam Wall (4X)

332

342

352

0 100 200 300Angle (degree)

Wa

ll T

hk

n

(um

)

Non-Concentricity: 2.7 +/- 0.2 %

5) Unflatten and record data4X lens radius measurement repeatability: <0.4 m

CH Wall (4X)

8

10

12

14

0 100 200 300

Angle (degree)

Wal

l Thk

n (u

m)

Wall Variation: 1.1 +/- 0.4 um

Correcting Walls

6) Apply offset correction (under development) •Measured profile has width

•Relation fixed between profile and interface•Specific to shell type and lens•Described by markers (peak/valley) and offsets

•Offset understood by modeling•Affected by phase contrast, pixel size, X-ray spectrum etc.

Fresnel calculation of 15umGDP/20umBe @ 10X

0

0.5

1

975 980 985 990 995 1000 1005 1010 1015 1020 1025 1030

Radius (um)

Am

plitu

de

-0.4

-0.2

0

0.2

0.4

Transmission

2nd Derivative

Be/GDP interfaceGDP surface Be surface

Porosity may affect interface sharpness

• Phase contrast shows as white ring at the sharp interfaces of dissimilar materials– Strong at CH/RF foam, CH/Be interfaces – but not for CH/DVB –Diffused due to pore size?

CH on DVB foamCH on RF Foam

~0.1 m pore ~1 m pore

50 m

50 m

Estimated X-Radiography Capabilities

Lens

Image resolution (m)

Maxim OD(mm)

MinimResolvable Layer (m)

Profile Repeat-ability (m)

Wall thkn Accuracy (m) *

Method

2X 3.7 6.0 10 0.8 1 2nd Deriv.

4X 1.9 3.3 6 0.4 0.5 2nd Deriv.

10X 0.8 1.4 2 N/A 1 Transmission

20X 0.4 0.7 1 N/A 0.5 Transmission

* After applying offsets determined by Fresnel calculation

Special concerns for DVB foam shells

• Use low mag. 2nd derivative analysis for foam– May not resolve thin CH overcoat– But get the complete foam radius profile

• Determines foam wall, shell diameter, OOR and NC

• Use high mag. transmission analysis for CH– Noise too high for 2nd derivative method– Measure local CH wall thickness

• Do NOT apply the offsets for Be/GDP shells– Offsets specific to shell type and shell size

Future Possibilities

• Noise analysis could give quantitative opacity variation– IFE specification: <0.3% density variation over

50-100um• No characterization method yet

– Calculate sample opacity from film transmission• Film model already developed for the ICF program

• Orthogonal views allow 3D NC measurement– 90˚ Rotating device

• While sample stays in XRF holder