Fatigue strength of metal sandwich type construction

• Main Author: Lamb, William E.
• Format: Others
• Published: 1949
• Online Access: https://thesis.library.caltech.edu/393/1/Lamb_we_1949.pdf; Lamb, William E. (1949) Fatigue strength of metal sandwich type construction. Engineer's thesis, California Institute of Technology. doi:10.7907/YEGJ-XT59. https://resolver.caltech.edu/CaltechETD:etd-01282009-095113 <https://resolver.caltech.edu/CaltechETD:etd-01282009-095113>

An investigation was conducted to determine the fatigue strength of a new type of all metal sandwich panel structure, designed and furnished by Western Engineering Associates of Los Angeles. This structure consisted basically of a new type of embossed core attached by spot welding to one or more smooth sheets of the same metal. Specimens tested were all of the single core, single skin type. Metal used was 24ST aluminum alloy of 0.032 inch thickness. This construction affords high structural rigidity for its weight. Two patterns of the embossed core were tested. One of the specimens had its core embossed with a triangular pattern, the other with a square pattern. The specimens were tested by utilizing them as simply supported beams, loaded in the center by a sinusoidally varying load of constant maximum magnitude. This load was applied by means of loading bars one on the top of the specimen, the other on the bottom. These bars had flat, one inch wide surfaces, contacting the specimen. To prevent the sharp edges of the bars from causing local failures of the specimens, a layer of one-eighth inch thick koroseal was used between the bars and the specimen. Specimens tested had a length between end supports of 16.7 inches. Their width was 9 inches. The loading obtained was a combination of bending and shear. The shear stress was of such a low magnitude, however, that it could be neglected. Failure of the specimens was deemed as occurring at the time the first crack appeared. A method was devised for crack detection that consisted of laying down a conducting strip over a thin insulating layer in a network fashion, covering all saddle points of the core in the central area, since previous testing had disclosed the fact that failure occurred at these points first. Any crack in one of these saddles caused a break in the conducting strip which changed the bias on the controlling electronic tube to a cut off value. This tube was part of an Eccles-Jordan Trigger circuit which with associated tubes allowed current flow through a thyratron relay circuit opening the starting circuit of the testing machine, causing it to stop. The data shows that for the specimens oriented in a normal fashion, that is, with one of the sides of the square or triangle of the core parallel to the loading bar, which was the situation for most of the tests, the square pattern is vastly superior to the triangular one, as regards fatigue strength. For specimens oriented in this way the square pattern withstood a bending moment of 13.45 inch pounds per inch of width; whereas the triangular pattern withstood only 9.28. Two tests conducted with the square core having the sides of the squares at an angle of forty-five degrees to the loading bars gave results about midway between. However, the effective EI for this configuration was considerably reduced, and consequently actual failure stress was probably about equal to that for the ease where the sides were parallel to the loading bar. The triangular pattern was actually much worse than the curve shows. At the higher loads cracks occurred with very few cycles of loading. Automatic cut-off feature was not in use for these tests. In service it is highly possible that overloads of short duration might cause small cracks which would become focal points of fatigue failure, thus reducing the fatigue strength well below the design point. This weakness of the triangular pattern arises from the fact that smooth fillets or saddles, joining depressed and elevated portions of the core, are harder to obtain in this pattern than in the square one. Wrinkles and tool marks were present in almost every one of the triangular specimens tested.