Pretest Prediction Composite Hat

A Large, All-Composite Fuselage Test Article

Summary

HyperSizer was recently applied to an approximately $2.5 million structural test of a space launch, all composite cylindrical shell. The article was 10 ft. by 12 ft. with three J shaped interior ringframes and 43 hat shaped stiffeners. The test took place August 1997 at NASA Langley Research Center. The test article was fabricated to be representative of an X33/RLV fuselage and was loaded in longitudinal compression.

The test article's purpose was primarily to investigate panel buckling. However, HyperSizer pretest analyses predicted that local buckling of the total facesheet (facesheet over and between the hats) would occur first around 2800 (#/in), preventing the shell from going into a panel buckling mode. Unfortunately due to a fabrication problems in the way the hats stiffeners were bonded to the facesheet, they pulled off during the test prematurely at a load of around 2000 (#/in) causing catastrophic failure. Therefore, neither panel buckling nor local buckling were attainted.

Important Observations About the Test Correlation Process

The general approach used by industry and NASA is to build different FEMs of the test article to capture different structural responses and potential failures. For this test, at least four distinctive (non-HyperSizer related) FEMs were built by the aerospace company and NASA, and a fifth considered but not made due to excessive computational requirements.

• Traditional analytical approach requires many different FEMs »
• HyperSizer requires only one, coarsely meshed, 2-D FEM »

Once the model’s connectivity, loading, and boundary conditions were complete, HyperSizer was used to generate the FEM material and property data including beam element offset vectors. This data was written by HyperSizer in native finite element format and was used by the modeler during the FEM building and checkout process, saving the effort of having to hand compute the stiffened panel and ringframe stiffness terms. Once good model properties are available, then the FEA was executed to obtain the correct internal element loads.

The setup of the finite element model coupled with HyperSizer is typical of any other loads model. It is defined by the following steps:

• Generate the planar 2-D quad shell and line beam elements to represent the RLV hat stiffened panels and J ringframes respectively
• Define the material orientation axis for the shell elements and an orientation vector for the beam elements
• Identify element zones which are to have the same property definition (Component definitions defined by PSHELL and PBAR records)
• Define initial properties for the elements
• Assign boundary conditions and applied external loads
• Run the Finite element analysis

Conclusion

HyperSizer's pre-test prediction was that local buckling of the facesheet would be the critical failure mode and would occur at a load of 2800 (#/in). Two independently made discretely meshed 3-D FEMs, (one by NASA) confirmed local buckling of the facesheet to be around 2800 (#/in) to 2900 (#/in).

Analytical failure correlations to the test were not possible due to the manufacturing flaw, though it is important to note that HyperSizer was able to quickly identify a critical failure mode of the structure so test engineers could focus efforts on a specific FEM that addressed this concern.