The performance in fire of restrained columns in steel-framed construction

• Main Author: Shepherd, Paul
• Published: University of Sheffield 1999
• Subjects: 624.1; Axial restraint
• Online Access: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.310890

The University of Sheffield has a non-linear finite element analysis program called VULCAN which has been developed in-house over a number of years and has been thoroughly validated. A parametric study has been conducted using this software, which assesses the influence of such factors as load, geometric imperfection, material properties, temperature profile and axial and rotational restraint on the behaviour, of isolated steel columns in fire. This study is then extended to investigate the behaviour of steel columns as part of a larger multi-storey frame, in which axial restraint to thermal expansion of the heated column is provided by the floors above. A method of modelling these effects in VULCAN using a linear spring element to provide axial restraint has been developed and validated. An experimental partner project has been carried out at the University of Ulster, in which steel columns were furnace-tested with various levels of load, slenderness and axial restraint. These tests have been paralleled with analyses using VULCAN and a good correlation with test results has been shown. The VULCAN program was then used to examine the effects of parameters outside the range of the physical constraints imposed by the test facility. A numerical model, capable of assessing the level of axial restraint imparted on a column by a general multi-storey framed structure has been developed, which has a number of levels of complexity, each giving a higher degree of accuracy. Once the level of restraint for a structure has been assessed, the parametric studies and test data can be applied, and conclusions drawn about the behaviour of the frame. The applicability of different mathematical solution procedures to the analysis of these columns, which exhibit snap-through and snap-back behaviour, has been conducted. The arc-length method has been identified as applicable in such cases and a skeleton version of the procedure introduced into the VULCAN program. The program structure of VULCAN has been improved and the format for data input and output has been developed to allow flexibility. A graphical file-viewer program has also been created. Details of these changes are shown in appendices.