Behaviour of cellular beams having different opening sizes, steel decking shapes and restraint effect in fire conditions

  • El-Hadi Ali Naili

Student thesis: Doctoral Thesis


The growing popularity of the use of cellular steel beams in composite floors comes at the same time as an increasing attention to fire safety engineering design. The recommendation for their design in fire limit states remains very primitive and this is due to the lack of general research in this area. A total of six simply-supported and restrained composite cellular floor beams subjected to elevated temperatures, were tested at The University of Ulster. The first three beams conducted were simply-supported of 4.5 m span lengths, with main variables, different steel geometries and opening shapes. The beams were subjected to single and two point loadings represented by load ratio of 0.3 determined from the pre-design and finite element modelling of the sections at ambient temperature for the evaluation of the failure loads. Deflections were recorded using linear variable differential transducers (L VDT) during the fire tests, and temperatures were measured using thermocouples located along the length of the steel beams and the composite slabs. The main failure modes in two tests were the web post buckling associated with Vierendeel bending; whereas the latter was the main failure mode in the symmetric composite cellular beam with two elongated web openings. Similar failure modes were observed in the finite element modelling at ambient temperature. The experimental results have been compared against the results obtained by DIANA software for finite element modelling and were in good agreement. The capability of the model demonstrated excellent prediction for the three tests in terms of temperature distribution, deflection behaviour and failure mode after improvements were made to the models. A parametric study has been conducted using the analytical based web-post failure characterized by the failure of the critical section for the evaluation of temperature and time of failure. The results were also compared with the experimental data and were found in good correlation when using the appropriate effective length subjected to buckling, which is different from one beam to another beam. However, it was found that the formula of shear buckling capacity of the web post at elevated temperatures provided in the SCI [22] documentation, which is expressed in terms of longitudinal shear needs to be adjusted in order to predict accurate effective length of post web buckling for cellular beams, with different geometries and cross sectional dimensions. The second phase of fire tests was restrained cellular beams with longer span lengths of 7.5 m. Different parameters were investigated by varying the depths of composite slabs and shape of steel decking compared to the tests of simply-supported beams. The beams were also pre-designed first at ambient temperature in order to evaluate the failure loading and 0.3 of this load was considered in the three tests. The finite element modelling at cold temperature showed that the beams considered as simply-supported failed due to flexural bending accompanied with Vierendeel mechanism at different load levels. Similar instruments and equipment have been used as in the tests of simply-supported beams, and the difference is represented by the restraint frame that was positioned around the furnace where the specimens were axially restrained and subjected to heating. In addition, strain gauges were used at different locations in the cold-ends of the cellular beams, top of the slabs and the frame for the measurement of strains. The failure modes in the tests were flexural bending associated with initial stage of Vierendeel bending and the buckling of the web posts. There was no occurrence of lateral buckling as expected due to the low degree of restraint and type of end-condition used. The web post buckling was the main failure mode in composite beam with the smallest depth of composite slab, a sudden change occurred at the last stage when compressive forces changed into tensile force that can lead to the development of catenary action and the occurrence of lateral buckling of the bottom flange. The effect of restraint on cellular beams has been carried out first using finite element mode ling before the fire tests where asymmetric and symmetric composite cellular beams with circular web-openings have been studied under high and low restraint factors at ambient and elevated temperatures. At ambient temperature, the main failure modes were web-posts buckling for both sections in case of low restraint factor as in the case of simply-supported beams; whereas web-posts buckling occurred and followed by lateral torsional buckling in case of high restraint factor. In both sections, compressive forces increase with the increase in loading, and the load-carrying capacity is significant in case of high restraint factor compared to low restraint factor due to high compressive force developed from composite slab and the catenary action allowing the section to experience maximum deflection before failure. At elevated temperature, the web-posts buckling was the main failure mode for a low degree of restraint; whereas, a severe web-posts buckling followed by lateral torsional buckling were the main failure modes for a high restraint factor. The restrained beams tested have been modelled using the model developed and all the data gathered during the tests in terms of temperatures recordings and axial stiffness. The model showed a good correlation with the tests results and was next extended beyond the limit of the experimental fire tests by considering the full fire exposed lengths of the cellular beams; by investigating the influence of different degrees of restraint conditions and depths of composite slabs having different shapes of steel decking.
Date of Award2012
Original languageEnglish
SupervisorAli Nadjai (Supervisor), Faris Ali (Supervisor), Seng-Kwan Choi (Supervisor), Seng-Kwan Choi (Supervisor), Ali Nadjai (Supervisor) & Faris Ali (Supervisor)

Cite this