Solutions
for Characterizing Coated Particle Nuclear Fuel
Submitted
by: Jeffery Price & Thomas Karnowski, Image Science and Machine
Vision Group
Introduction
The DOE Advanced Gas Reactor (AGR) Fuel Development and Qualification
program was conceived to support near-term deployment of high-temperature,
gas-cooled reactor technology and to establish a basis for the development
of fuels suitable for very high-temperature, gas-cooled reactors. The
first phase of this program concentrates on the re-establishment and improvement
of the capability to produce high quality, silicon carbide-based, coated
fuel particles (also known as ceramic-coated or TRISO fuel). A cross-section
representation of these spherical particles is shown in Fig. 1. A new
coated particle fuel development and characterization facility is being
established at ORNL. Characterization is required to support coating development
work and predict the ultimate performance of the fuel particles. Technological
advances, particularly in the areas of digital imaging and computing,
offer the opportunity to improve upon the characterization methods that
were developed two to four decades ago when coated particle fuel was first
developed. Members of the Image Science and Machine Vision Group have
developed solutions to enhance the efficiency and accuracy of the coating
characterization.
 |
| Figure 1. Cross-sectional view of the (nominally) spherical coated particle fuel. Each complete particle is composed of five layers. |
Light Microscopy and
Image Processing
One portion of the particle characterization effort involves light microscopy
and image processing. The particles to be inspected are imaged in two
different phases. In the first phase, particles or kernels are collected,
in a single layer, in an optically transparent dish and imaged using back-lighting.
The resulting images, an example of which can be seen in Fig. 2, are analyzed
to measure the outer surface characteristics only, as noted in Fig. 3.
In the second phase of optical inspection, complete particles are embedded
in an acrylic epoxy and ground down to just short of the spherical midpoint
to produce images such as that in Fig. 3. These images are used to measure
width characteristics of all the material layers in the particle as noted
in Figs. 4 and 5.
 |
| Figure 2. Multiple particles imaged with back-lighting for outer surface characterization. |
 |
| Figure 3. Outer surface characterization. Distinct particles are identified from images such as that in Fig. 2 and then their shapes are characterized and recorded. Shape characterization includes measuring the radii of the particle at 360 points (the red boundary), the maximum and minimum radii (red and blue solid lines, respectively), the maximum and minimum diameter (red and blue dashed lines, respectively), and the maximum and minimum curvature (red and blue triangles, respectively). All processing is done automatically. The analysis program runs in a service-like mode on a compute server; new images are analyzed when they are copied to the watched input folder. |
 |
| Figure 4. An example image used for cross-sectional layer thickness measurements. The dark (shadow) region just outside the fifth layer is the excess particle resulting from not being able to grind the particle down to the true center. The thickness of the shadow region is used to compute how far from true center the observed cross-section is and, furthermore, to correct the layer thickness for this offset. |
 |
| Figure 5. The particle center is estimated and then the image is “unwrapped” by sampling on a polar grid. Vertical scale represents angle in degrees and horizontal scale represents microns. |
 |
| Figure 6. Boundary-finding methods are employed to locate the layer edges in the unwrapped image. Layer thicknesses are then measured and recorded. Again, all processing is done automatically. |
TRISO Particle
Counting and Sizing Tool (TP-CAST)
The TP-CAST was developed to address the need for rapid counting
and measurement. The developed system, illustrated in Fig. 7, employs
a light obstruction concept. The TP-CAST consists of an optical system
that projects light through a target transport cell and collects the light
onto a photo-receiver. The signal from the photo receiver is digitized
by a high-speed ADC unit and is processed in real-time. The events are
counted and each event is analyzed to estimate the size of the particle.
The particles are pulled via a vacuum through a small piece of tygon tubing,
shot through the target cell which is fitted with an optical window, and
collected with a cyclonic separator.
 |
| Figure 7. Diagram of TRISO Particle Counting and Sizing Tool (TP-CAST). |
The TP-CAST was developed for high-speed counting and size measurement of TRISO spheres of diameter from 300 to 1000 microns. Counting accuracy was estimated as less than 0.075% error with a 95% confidence. The TP-CAST size measurement accuracy is on the order of 11 microns standard deviation for spheres 1000 microns in diameter. Although the current particle transport system does not support the maximum detection rate, electronically generated data showed rates of 200 particles per second, implying a potential throughput of 720,000 particles per hour.
 |
 |
| Figure 9. Completed TP-CAST instrument in light-shielding enclosure. (Designed by Curt Maxey; Advanced Lasers, Optics, & Diagnostics Technology Group.) |
References
J. Price, J. Hunn. “Optical Inspection of Coated Particle Nuclear
Fuel,” in Proceedings of Machine Vision Applications in Industrial
Inspection XII, SPIE vol. 5303, pp. 137-149, January 2004.
T. Karnowski, A. Kercher, J. Hunn, C. Maxey. “A Simple Optical System for Real-time Size Measurements of TRISO Fuel Pellets,” in Proceedings of Machine Vision Applications in Industrial Inspection XIII, January 2005 (to appear).