Experimental and numerical analysis of GFRP frame structures. Part 1: Cyclic behaviour at the connection level

Abstract

Pultruded glass fibre reinforced polymer (GFRP) profiles have low weight, high strength and corrosion resistance, but their brittle failure raises concerns about their use in seismic regions. Moreover, although their static monotonic response is reasonably well understood, the cyclic and hysteretic behaviour of GFRP frame structures and their beam-to-column connections have not yet been comprehensively investigated. This paper presents experimental and numerical investigations on the cyclic behaviour of a novel tubular GFRP beam-to-column sleeve connection system, comprising internal metallic parts. Four series of the connection system were tested, with varying number and position of the beam connection bolts, namely with: (i) one bolt in the webs (W1); (ii) two bolts in the flanges (F2); (iii) four bolts in the flanges (F4); and (iv) two bolts in the flanges with larger edge distance (F2S). The results show that series W1 presents the worst overall cyclic behaviour. On the other hand, the addition of bolt rows (F4 vs. F2) did not improve the cyclic response of the connection system. Conversely, increasing the edge distance (F2S vs. F2) led to significant improvements of the hysteretic behaviour, namely in the capacity to dissipate energy. Using the Pivot hysteresis model in the numerical study, a design-oriented model comprising frame and link elements was developed to simulate the response of the best performing series (F2S). In spite of its simplicity, the numerical simulation provided good agreement with the experimental results. In a companion paper, the behaviour of full-scale frames comprising F2S connections under monotonic and cyclic quasi-static sway tests was experimentally assessed. Numerical models of these tests were also developed to simulate the cyclic behaviour of the frames, using the parameters of the Pivot hysteresis model calibrated herein.