Tbe study consists of an experimental and analytical investigation into the lateral buckling behaviour of steel I-section girders braced by continuous or discrete
U-frames. Scaled down laboratory tests on twin I-section girders have been carried out under full instrumentation and are reported. Lateral deflection of the compression
flanges and final buckling modes were recorded and the coupling effect of U-frame action is clearly demonstrated. The failure loads obtained were generally higher than
the corresponding design values according to BS5400.
Using a large displacement elasto-plastic finite element package, ABAQUS, finite element idealisations of the tests were established and analysed. Good correlation between the experiments and the numerical analyses was reached. Validity
of the ABAQUS package was confirmed and first order elements were sufficiently effective for the analysis.
This was followed by further investigation of a wider range of I-section girders using ABAQUS. The ultimate bending resistance of the girders obtained from finite element analysis was in general greater than the corresponding design values to BS5400, particularly so with girders of high slendemess. Thus the present design method is considered to be unduly conservative.
The cause of this conservatism in BS5400 is discussed. The expression for the calculation of effective length of U-frame braced girders was found to be reasonable. Based on the results from ABAQUS and reviews of the limited research directly related to this study, two main parameters in the BS5400 expression for beam slenderness seem inappropriate for lateral buckling of an I-section girder under U-frame restraint. These are the radius of gyration of the whole girder section and the ratio of overall depth of a girder to mean flange thickness. Instead, radius of gyration
of the compression flange together with a contribution from the adjoining web section was found to be more appropriate as also was the web slenderness. In addition, a modification is proposed to the present limiting stress curve for lateral buckling of bare steel I-section girders. Empirically derived factors have been introduced. The
general procedures in the existing design method to BS5400 remain similar, however.