Issue #14 - April 28th, 2008
Software Suite for Durability, Damage Tolerance, Reliability & Life Prediction
Enhances FEA Solvers MSC Nastran*, ABAQUS, ANSYS, RADIOSS & LS-DYNA
* Best Performance and Verified Solutions with MSC Nastran
This Week's Feature Composite Example
Composite Storage Module Joint Analysis and Test Verification
|
The premature failure of adhesively bonded composite joint tests, suggest adhesive failure as opposed to cohesive failure. This type of joint failure may be due to a) non-clean interface surfaces preparation, b) thick bond line, c) existence of voids in the adhesive bond line or a combination of all three
In support of the ONR (Office of Naval Research) SBIR Phase II program, several Naval joint structures were studied. These joints included a General Dynamics designed joint for use on a composite storage module, and two test articles of a General Dynamics/Boeing designed joint attaching the walls of a director's room to the deck of a ship. During testing, the General Dynamics Composite Storage Module (CSM) joint failed prematurely and progressive failure analysis was employed to shed light on the possible causes of the premature failure.
The General Dynamics Composite Storage Module (CSM) joint was a particularly useful example structure as a thorough analysis and testing program that accompanied the design and verification of the joint development. A significant portion of this program was developed into a paper [1]. The CSM joint version of interest was composed of a graphite/epoxy quasi-isotropic laminate adhesively bonded to a tapered titanium fitting. The CSM and joint details are shown in Figure 1. Details of the FE model and materials used in the analysis are shown in Figure 2. The tensile load on the joint bends the joint resulting in peel stresses in the adhesive EA9394 layer. The test indicated that the primary failure mode of the joint was adhesive failure within the EA9394 layer between the titanium and fiberglass.
 |
Figure 2 - FE Model Details and Materials for Progressive failure Analysis (PFA) [1]
|
Several steps were taken before the final GENOA Progressive Failure Analysis (PFA) was performed in order to develop and demonstrate confidence in the software. For example, as a baseline check, linear analyses were performed using standard FEA software packages ANSYS and ABAQUS along with GENOA. General Dynamics performed the ABAQUS simulation. Stress results along the adhesive centerline of the EA9394 adhesive are shown in Figure 3 for all three codes. The validation of GENOA simulation to capture the nonlinear material behavior of the adhesive was achieved by duplicating the ABAQUS analysis performed by General Dynamics.
 |
Figure 3 - Comparison of Centerline Peel Stress Predictions Within the (Linear) EA9394 [1]
|
A full PFA with GENOA was then performed and compared to the test data. The predicted load deflection curve and associated damage mechanisms are shown in Figures 4a and b. While the failure mechanisms predicted by the analysis were very similar to those observed during the test, the analysis predicted a much higher strength for the joint as shown in Figure 4a. The tested joint may have failed at a much lower load (see blue curve in Figure 4a) because (1) the titanium surfaces were improperly prepared for the adhesive and/or (2) the maximum value of strain the adhesive could withstand was lower than that assumed from measurements of the adhesive done in 1995 and provided in reference 3. Without some relatively inexpensive coupon-level tests to establish a more relevant value of adhesive strength/strain limit, the failure strain criterion was not adjusted. However, improper surface preparation was investigated further.
Subsequent review of the failure surfaces revealed a clean titanium surface indicating a premature interfacial failure of the adhesive to titanium bond (Figure 1). Analyses were performed to introduce the clean surface interfaces on the titanium fitting (test suggest adhesive failure as opposed to cohesive failure) using: 1) Virtual Crack Closure Technique (VCCT), and 2) using PFA and degrading the adhesive properties assuming using 20% void formation [2].
VCCT required a predetermined crack path, and was modeled in two ways; a) multiple crack locations, and (b) single crack initiation point. For these simulations the crack path was determined from PFA (Figures 4b and 5a). The results obtained predicted higher strength value than predicted by PFA strain based analysis (blue curve in Figure 4a). Plane strain fracture toughness for peel and shear, (K1C and KIIC), values were required to run the analysis using the VCCT approach. The K1C and KIIC values for EA9394 adhesive were obtained from somewhere else [3]. Note that PFA predictions are usually close but are not considered very accurate when pre-cracked conditions exist in a simulation due to infinite stresses near a sharp crack tip.
 |
Figure 4 - Load-deflection curve and associated damage events [2]
|
Thereafter to capture the effect of improper bonding, the VCCT analysis was defined on limited node pairs as shown in Figure 5b. The trend of the results indicated that defining the partial bonding can simulate the effect of improper bonding, as shown in Figure 6.
Similarly a void content of 20% was used to simulate the improper bonding in PFA. Again the results were found to be a close match with that of the test (see Figure 6).
 |
Figure 5 - (a) Single VCCT approach. Note that the crack initiates from the right hand side and propagates to the left. (b) Partial Single VCCT approach to simulate partial (improper) bonding. The left hand side images are the initial setup and the right hand side is the final failure of the bonding [2].
|
 |
Figure 6 - Comparison of the simulated load displacement curve with test data [2].
|
In conclusion, if proper contact surface area information is available before the analysis, PFA and VCCT can be used judiciously to simulate the effect of improper bonding. Moreover, insight into potential improper bonding can can be achieved by varying the voids percentage in the adhesive to simulate the improper bonding scenario if surface area is not available.
References:
1. George F. Leon, Michael F. Trezza, Jeffrey C. Hall,1 and Kelli Bittick, "Evaluation of a Carbon Thermoplastic to Titanium Bonded Joint", ASTM-STP1455-11707.081503. Click here to read technical publication.
2. Xie, D., Garg, M., Huang, D., and Abdi, F., "Cohesive Zone Model for Surface Cracks using Finite Element Analysis," AIAA-49SDM-106742-2008. Click here to read technical publication.
3. T. R. Guess, E. D. Reedy, M. E. Stavig, 1995. "Mechanical Properties of Hysol EA-9394 Structural Adhesive," SANDIA REPORT, SAND95-0229. UC-704. Click here to read technical publication.
This issue was brought to you by Alpha STAR Corporation.
If you do not want to receive this newsletter, please click here. If you still have trouble unsubscribing from our website, please email genoa@ascgenoa.com with a "UNSUBSCRIBE" subject header.
If you do not want to receive this newsletter, please click here. If you still have trouble unsubscribing from our website, please email genoa@ascgenoa.com with a "UNSUBSCRIBE" subject header.
No comments:
Post a Comment