Experimental investigation of shear connector behaviour in composite beams with metal decking

Presented are experimental results from 24 full-scale push test specimens to study the behaviour of composite beams with trapezoidal profiled sheeting laid transverse to the beam axis. The tests use a single-sided horizontal push test setup and are divided into two series. First series contained shear loading only and the second had normal load besides shear load. Four parameters are studied: the effect of wire mesh position and number of its layers, placing a reinforcing bar at the bottom flange of the deck, normal load and its position, and shear stud layout. The results indicate that positioning mesh on top of the deck flange or 30 mm from top of the concrete slab does not affect the stud's strength and ductility. Thus, existing industry practice of locating the mesh at a nominal cover from top of the concrete slab and Eurocode 4 requirement of placing mesh 30 mm below the stud's head are both acceptable. Double mesh layer resulted in 17% increase in stud strength for push tests with single stud per rib. Placing a T16 bar at the bottom of the deck rib did not affect shear stud behaviour. The normal load resulted in 40% and 23% increase in stud strength for single and double studs per rib. Use of studs only in the middle three ribs out of five increased the strength by 23% compared to the layout with studs in first four ribs. Eurocode 4 and Johnson and Yuan equations predicted well the stud strength for single stud/rib tests without normal load, with estimations within 10% of the characteristic experimental load. These equations highly under-estimated the stud capacity, by about 40-50%, for tests with normal load. AISC 360-16 generally over-estimated the stud capacity, except for single stud/rib push tests with normal load. Nellinger equations precisely predicted the stud resistance for push tests with normal load, with ratio of experimental over predicted load as 0.99 and coefficient of variation of about 8%. But, Nellinger method over-estimated the stud capacity by about 20% in push tests with single studs without normal load.