Experimental Seismic Performance of Partly-Sliding Partition Walls

Abstract

ABSTRACT Plasterboard partition walls typically used in commercial buildings are especially sensitive to earthquakes, with the onset of cosmetic damage initiating at small values of interstorey drift. The most common partition wall systems are constructed of gypsum board attached to either steel or timber framing which is fixed directly to the floor system at the top and bottom interfaces. This study investigates the seismic performance of a novel partly-sliding steel-framed partition system examined in the past and used by industry, with minor modifications incorporated within the partition detailing. This novel system involves removing top track anchors within the proximity of wall intersections, thus allowing the tracks to ‘bow’ out at these locations. In this study three full-scale specimens were subjected to quasi-static cyclic testing; two identical plane specimens and the third including a doorway. The specimens were built in a y-shape and angled at 30° to the direction of applied loading, which allowed bi-direction behaviour to be examined. The specimens included an acoustic/fire sealant. The progression of damage in a partition can be categorized by three sequential damage states associated with distinct levels of repair: superficial damage requiring cosmetic repair (damage state 1 (DS1)), damage requiring local repairs or replacement of only portions of the partition assembly (damage state 2 (DS2)), and severe damage requiring complete removal and replacement of the wall (damage state 3 (DS3)). Damage was first observed as cracking of the wallboard at the wall ends, at external junctions, and propagating from the corners of the door opening. The onset of DS1 and DS2 occurred simultaneously at a median in-plane drift of 0.29%. DS3 was not observable until the linings had been removed at the end of the tests. In addition to providing drift capacities, the force-displacement behaviour is also reported, the dissipated energy was computed, and the parameters of the Wayne–Stewart hysteretic model were fitted to the results. The specimen with the door opening behaved significantly different to the plane specimens: damage to the doorway specimen began as cracking of the wallboard propagating from the corners of the doorway following which the L- and Y-shaped junctions behaved independently, whereas damage to the plane specimens began as cracking of the wallboard at the top of the L-junction and wall system deformed as a single unit. The results suggest that bi-directional behaviour is important even if its impact cannot be directly quantified by this experiment. Damage to sealant implies that the bond between plasterboard and sealant is important for its seismic performance, and careful quality control is advised, as defects in the bond may significantly impact its ability to withstand seismic movement.