On plate buckling coefficients

Though thin plates have a useful reserve of elastic postbuckling strength, the linear buckling load remains an important design parameter. White and Cottingham have published useful graphs of buckling coefficients versus aspect ratio for various partial edge loadings and for simply-supported as well as for fixed-edge boundaries. Some of their graphs exhibit large cusps which they attribute to a change of buckling mode. The writers have written a fairly general finite-difference plate buckling program in preparation for postbuckling studies; to test the program they recalculated the White and Cottingham results. These recalculations not only removed all cusplike buckling coefficient response for fixed-edge boundaries but, for one loading case, found that the White and Cottingham coefficients appear 100% too high whereas for another loading case the writers' results were 1/8 higher; for simply-supported boundaries the results agree closely. The writers' conjecture is that the cusplike behavior found by White and Cottingham is due to erroneous generalized eigenvalue solutions in the previous calculations, removed by the writers' use of reliable EISPACK subroutines.