Evaluation of compression members with non-ideal end conditions

Abstract

This thesis presents the results of research in which a formulation was developed for the effective length factor for compression members with non-ideal end conditions. This research stems from previous research conducted by Kohutek, Nunn, and Faseler (1993), in which twelve tubular members were tested in axial compression. These members possessed non-ideal end conditions consisting of nontranslating ends and restraining moments. The test specimens were divided into four groups based upon the condition of the member (undamaged/ungrouted, damaged/ungrouted, undamaged/grouted, damaged/grouted). The author developed four different methods to measure the rotational restraint of the load frame used in the testing by Kohutek, Nunn, and Faseler (1993). The properties of the tubular members were calculated based on four different idealizations, also developed by the author. The effective length factor was found by using these rotational restraint measurements and specimen properties in a transcendental equation originally developed by Timoshenko and Gere (1961) for a beam-column with symmetric rotational restraints. The computed effective length factor based on the author's formulation and idealizations was compared with the experimental effective length factor determined by Kohutek, et al. . The computed effective length factor values were, in general, conservative predictions (i.e. KTMEO > KExp). For each group of tubular specimen, a particular method of computing the rotational restraint and the cross-sectional stiffness was identified by use of the lowest percent error between experimental K and theoretical computed K. The load frame used by Kohutek, et al., did not provide the same rotational restraint at each end of the specimen. Therefore, a modified rotational restraint was computed based on the average of the two end rotation values and was found to provide more accurate values of the effective length factor. For a specimen with an unknown effective length factor, a simple procedure has been developed to accurately predict the effective length factor, given certain cross-section data and full-scale testing data.