Date of Award

Spring 2005

Document Type

Thesis - Open Access

Degree Name

Master of Aerospace Engineering

Department

Graduate Studies

Committee Chair

Dr. Yi Zhao

Committee Member

Dr. Eric v. K. Hill

Committee Member

Dr. David J. Sypeck

Abstract

Composite materials have become one of the leading materials for manufacturing in the aerospace industry today. Compared to "conventional" aerospace metals, composites generally have higher strength-to-weight and stiffness-to-weight ratios, good fatigue and corrosion resistance and reduced parts count. However, like any other material, they also have disadvantages. They are inherently brittle and are thus prone to impact damage. Low energy/velocity impact damage, in particular, can be dangerous because the damage oftentimes goes undetected and can subsequently grow under load. Also known as barely visible impact damage (BVID), this area of concentration focuses on the small-scale damage that may be very difficult to detect yet can be lethal.

The primary emphasis of this research is to predict the residual compressive strength of a 16-ply laminate [(0°/±45°/90°)2]s after experiencing low energy/velocity impact using combined technical approaches of ultrasonic C-scan and neural networks. To accomplish this, each test specimen was ultrasonically C-scanned after impact testing. A MATLAB computer program was then used to convert the image files into numeric data, which they were presented to a backpropagation neural network in order to predict the residual compressive strength. Microsoft Excel was used to take the average of the diagonal values of the normalized image data. Here the average prediction error turned out to be 3.9 percent, while the worst-case prediction error was 14.6 percent.

This research also focused on identifying, sorting, and classifying how the composite laminates failed under compression after experiencing low energy/low velocity impact. Acoustic emission (AE) parameter data were collected during compression testing, and then inputted into an artificial neural network (ANN) for classification. Specifically a Kohonen Self Organizing Map (SOM) was used to sort and classify the failure mechanisms that occurred within the weakened composites. The associated BVID failure modes, otherwise known as failure mechanisms, were believed to consist primarily of transverse and longitudinal matrix cracks, delaminations, and occasionally fiber breaks. Even though delaminations are the most critical failure modes in BVID under compression, the other failure mechanisms also contribute significantly. Furthermore, it appeared that it was also possible to sort out and determine the transition regions between BVID and visible impact damage (VID) with AE data. Thus, it is important to know how the material fails so that necessary precautions can be taken to minimize these critical failure modes.

Share

COinS