Chapter 3 Detecting Matrix Cracking in Composites
In the previous chapter, wave propagation is utilized to add robustness to the impedance method for structural health monitoring. The concept of the impedance method is introduced, but Chapter 2 only deals with how temperature influences wave propagation and methods to correct these temperature variations. In this chapter and the following chapter, the impedance method is used by itself to detect damage. Specifically, the amount of damage in composite test samples is quantified using impedance-based structural health monitoring.
The ability to detect transverse matrix cracking in composites can be very helpful in many applications where composites are used. In a [0m/90n]s setup, there are m layers of laminate with fibers aligned in a uniform direction of the outside, and n layers of laminate perpendicular on the inside. The total lay-up is symmetric (denoted by s), so there are n more layers of 90° in the middle, and m more layers of 0° on the other side. If a tensional load is applied along the fiber direction of the 0° layers, the first sign of the composite failing will be cracks in the 90° layers (Hyer, 1998). These cracks will appear parallel to the fibers in the 90° direction. The presence of these cracks causes larger stresses in the 0° due to stress concentrations. The stress concentrations and larger stresses lead the laminate failing in the 0° direction at a lower load than would occur if the 90° layers were not present (Hyer, 1998). The same type of cracking has also been shown to occur with fatigue as well (Kander, 1991). Thus, detecting matrix cracking can be essential to verifying the condition of a structure.
Many methods have been implemented in an attempt to detect transverse matrix cracking. Some of these methods include microscopy, edge replication, X-ray techniques using a dye penetrant, backscatter ultrasonics, and acoustic emissions (Kander, 1991, Prosser et al., 1995, and Subramanian et al., 1994). However, these techniques, with the exception of acoustic emissions, are not useful for in situ monitoring of a structure (Prosser et al., 1995). Even during testing, specimens must be taken out of loading conditions to detect cracking. Using penetrant to perform an X-ray will cause permanent damage to the composite, making further testing impossible. Therefore, a reliable method of detecting matrix cracking, whether a sample is being tested in the laboratory, or a real world structure is being used, needs to be able to detect cracking without disrupting the test or normal use.
The impedance method for structural health monitoring utilizes PZT patches as self-sensing actuators, as described in the introduction to Chapter 2 (Park et al., 2003). When the PZT is mounted on a structure, it can be used to actuate a structure at high frequencies. Since the PZT is bonded to the structure, it can be shown that the mechanical impedance of the structure is correlated with the electrical impedance of the PZT (Liang, et. al., 1994). Thus, any change in the mechanical impedance, which could be caused by damage to the structure, will show up in the electrical impedance. Using impedance as a method of detecting damage has been shown effective for many types of structures (Park et al., 2003). Early testing has shown the impedance method to already be useful in structures where composites are commonly used, such as aerospace applications (Chaudhry et al., 1995).
As previously mentioned, acoustic emissions have been used to detect matrix cracking. Specifically, Prosser et al. performed acoustic emissions tests of graphite/epoxy composites with a [0m/90m]s stacking sequence where m ranged from 1 to 6 (Prosser et al., 1995). The results of the experiment showed an exact correlation between signals observed with acoustic emissions and actual matrix cracking for samples were m is greater that 2 (Prosser et al., 1995). However, for samples where m was less than 2, cracking was not always successfully detected (Prosser et al., 1995).
In this chapter, an initial experiment to determine whether or not the impedance-based structural health monitoring method could be used to detect transverse matrix cracking in thin, or cross-ply specimens with only four layers, is described. Composites are fabricated and prepared for testing. Test coupons are loaded in tension to specific values, and impedance testing, utilizing PZT bonded to the coupons, is performed before and after loading. The impedance signatures are then analyzed to show that matrix cracking can be detected using the impedance-based structural health monitoring method. The same method is then used to perform damage detection on a thicker cross-ply composite.
Test Setup and Experimentation
The material used in this study is Narmco/BASF 5245C/G40/600, which is a prepreg with unidirectional graphite fiber in a bismaleimide resin. Prepreg sheets of 30.48 cm square are arranged to form a stacking sequence of [0/90]s. With only two 90° layers, the laminate is considered thin using the definition set by Prosser, et al. After curing the laminates, the panels are cut into fourths in order to allow tabbing material to be applied. The tabs protect the composite from the tensile testing jaws and provide a good surface for the jaws to grip. With tabbed panels test coupons with a length of 147.9 mm along the 0° fiber direction and 25.45 mm wide are cut. The samples are 1.29 mm thick. PZT patches are mounted to the composite coupons allowing impedance measurements to be taken. A sample coupon can be seen in Figure 3.1.
To induce transverse matrix cracking in the specimens, a quasi-static tensile test is performed. Different samples are subjected to different loads using an Instron machine with a 50 kN load cell. The different loading conditions allow a different number of cracks to form in each coupon. The loading is stroke controlled at a rate of 1 mm per minute.
Before each sample is loaded, and out of the Instron machine jaws, two baseline impedance measurements are taken with an HP 4192A impedance analyzer. Two more impedance measurements are acquired after the samples are loaded, again out of the grips. Impedance frequency ranges of 10-20 kHz and 40-60 kHz are chosen to take measurements over. Generally, there are a large number of peaks in these frequency ranges to compare the before and after loading condition impedance curves. A sample impedance measurement can be seen in Figure 3.2.
As seen in Figure 3.2, the impedance signature after the sample has been loaded is significantly different from the baseline. These changes are what will be used to determine how much damage, seen in cracking, is present in the sample. It should be noted that these curves have been normalized in the vertical direction to reduce the effect of drift or other variations. Also, the lengths of specimens were measured before and after loading, and no residual strain was noticeable. The stress-strain curve for a sample composite specimen is shown in Figure 3.3.
The strain before failure looks fairly elastic, as Figure 3.3 reveals. Elastic deformations would account for the lack of any residual strain in specimens. The shifting of the curve near 0.012 mm/mm strain is due to large accumulations of fiber fracture.
After each sample is tested, they are soaked in a dye penetrant and X-rayed. The penetrant soaks into the cracks, allowing the cracks to appear in the X-rays. The X-ray is used as verification for the number of cracks present. The length of the sample in between the tabs is 7.54 cm. To get a standard for the damage in a sample, the total number of cracks in between tabs are counted, and divided by the length to get cracks per centimeter. An enlarged view of an X-ray showing cracking can be seen in Figure 3.4.
Now that the actual damage for each sample is known, the impedance curves can be analyzed to determine how much damage was observed using structural health monitoring. In order to determine the amount of damage, some sort of damage metric must be implemented. A damage metric gives a standard interpretation of the difference between two impedance curves. For these test, a simple variation Root Mean Square Deviation (RMSD) is applied. The RMSD compares individual peaks to see how much they changed. The RMSD method for finding the damage metric, M, can be described as where Zi,1 is the baseline, or healthy, impedance of the PZT, and Zi,2 is the impedance used for comparison with the baseline measurement at frequency interval i. For the RMSD, the higher the damage metric value, the more difference there is between the baseline impedance signature and the impedance signature indicating damage. More complicated damage metric algorithms are available which take into account temperature and other variables, but for this initial experiment, temperature is assumed to remain constant during testing in a laboratory environment (Park et al., 2003). A brief discussion of different statistical damage algorithms can be found in Section 4.4.
Appling the RMSD to the impedance curves for each sample, damage metrics for each specimen are calculated. The damage metrics are normalized with the damage metric for the sample that is loaded to 20 kN. A bar chart that compares the baselines to the damaged signatures can be used to show the damage metrics graphically. The resulting plot can be seen in Figure 3.5.
Chapter 1 Introduction
1.1 Structural Health Monitoring Challenges
1.3 Literature Review
1.4 Thesis Overview
Chapter 2 Temperature Influences on Wave Propagation
2.1 Introduction to SHM and the Impedance Method
2.2 Analysis for One Dimensional Structures
2.3 Discussion and Applications
Chapter 3 Detecting Matrix Cracking in Composites
3.2 Test Setup and Experimentation
3.4 Detecting Matrix Cracking in Thicker Specimens
Chapter 4 Detecting Delaminations in Composites
4.2 Test Setup and Experimentation
4.4 A Comparison of Damage Metrics
Chapter 5 Developing a Wireless System for Impedance-based SHM
5.1 Introduction to MASSpatch Project
5.2 MASSpatch Hardware Development
5.3 Extending the MASSpatch Capabilities
5.4 Chapter Summary
Chapter 6 Conclusions
6.1 Brief Thesis Summary
6.3 Future Work and Recommendations
GET THE COMPLETE PROJECT
Considerations of the Impedance Method, Wave Propagation, and Wireless Systems for Structural Health Monitoring