INTRODUCTION

Traditional fire design procedures consider structural behaviour in terms of the performance of the element rather than the system. This is a consequence of relying on standard tests on isolated structural elements. Little consideration has been given to member interaction or to the crucial role of connections in maintaining overall stability in the event of a fire.

In recent years substantial research programmes have focussed on whole building behaviour, particularly fire performance, stimulated by a number of events, including:

• the programme of full-scale fire tests at BRE's Cardington laboratories in the 1990s

• the terrorist attacks on the World Trade Center in September 2001 and the subsequent reports by US and UK experts

• the development of a more rational approach to connection design in fire as set out in the Structural Eurocodes

• collapses of multi-storey framed buildings in Madrid in 2005 and Delft in 2008.

Major earthquakes in urban areas have often been followed by large fires that have been difficult to control and have caused extensive damage to property. Seismic-induced fire is a scenario with a high probability of occurrence as evidenced by the earthquakes in Northridge, California and Kobe, Japan. This scenario should therefore be addressed in performance-based designs for buildings in seismic areas.

Fire and earthquake are accidental actions (as defined in the Eurocodes) and are generally treated independently. The research presented in this paper attempts to couple fire safety with seismic safety based on the design of partial strength composite connections in order to provide:

• seismic safety with respect to accidental actions

• fire safety where seismic actions cause the stiffness and strength of beam-to-column joints to be reduced.

Two types of connection have been developed as part of a collaborative European project to develop fundamental data, design procedures and promote ductile and fire resistant composite beam-to-column connections[6]. This paper deals principally with the large-scale tests undertaken at BRE. For a more thorough analysis of the thermal and mechanical behaviour the reader should consult reference 6. The first connection (Type 1) consists of a partially reinforced-concrete-encased H-section column. The second (Type 2) consists of a reinforced-concrete-filled circular hollow section column. For both types of specimen both internal and external connections have been considered. In all cases the connection has been to I-section composite beams. Two types of slab have been incorporated. The first is a conventional composite deck consisting of profiled steel sheeting fixed to the beam using shear studs. The second is a precast ‘biscuit' slab with electro-welded lattice girders providing reinforcement in the form of trusses. For the Type 2 specimens the precast units incorporated polystyrene fillers to reduce the overall weight of construction.

The BRE experimental programme aimed to evaluate the fire resistance of the connection following earthquake damage. The criteria for assessment was that the connection should provide 15 minutes fire resistance once damaged by the earthquake. Because of the difficulty of transporting fully assembled pre-damaged connections, the specimens were erected, cast and tested at BRE prior to conducting the fire tests. The experimental programme was therefore in two parts: a static load test to simulate the effects of an earthquake, followed by a fire test.