Imagine a scientific experiment with a name that is actually indicative of what the experiment is about -- that's exactly what the fiber pull-out test is. If you've ever yanked on your brother's or sister's hair, and separated hair from head, you've performed this test!

Since we scientists don't experiment on humans (graduate students as guinea pigs being the lone exception), we perform out tests on composites. The sample used is a small disk of the matrix material with the fiber embedded in it. The matrix is clamped in place while the fiber is pulled. The point of maximum load is reached just before the fiber debonds, after which the load corresponds, not to adhesion between the fiber and matrix, but to the frictional forces of the fiber being pulled out. In learing the force needed to debond the fiber from the matrix, we learn about the mechanism of failure and just what it takes (or doesn't take) to ruin a perfectly good composite.

The test above gives only one data point. To get more data we embed longer and longer sections of the fiber in the matrix and repeat the test. Once you have lots of data points, a plot of the embedded lengths vs. the measured force may be generated. The data points are usually scattered around, but a fitted line can be squeezed out of it. The slope of this line is the shear debonding stength.

It's important in the experiment to avoid breaking the fiber, but if the fiber won't pull out and does break, that's acceptable in application because it means you've got a good composite.

Information referenced from Advanced Polymer Composites: Principles and Applications, Bor Z. Jang, ASM International, 1994.