Progressive collapse of framed structures

Progressive collapse of buildings triggered by local damage, when due to gross design-construction mistakes or malicious terroristic attacks, can produce catastrophic scenarios involving large economical and human losses but is also a key feature of most of the controlled demolition processes involving blasts. Therefore, the progressive collapse of 4 bays times 4 bays times 4 storeys framed structures after the sudden loss of one column has been simulated by means of a Discrete Element algorithm. The structure is made of columns supporting horizontal grids of principal beams and thin ceiling slabs. The system is meshed with 5081 massive spheres connected in pairs by 9248 linear elastic Euler-Bernoulli (EB) 3D beam elements and is subjected to gravity acceleration and dead and live load. The generic EB beam is removed from the system (brittle rupture) when its bending and tensile strain are large enough to satisfy a coupled breaking criterion. Disconnected portions of structure can therefore fall colliding with still intact parts and the subsequent rigid inelastic impacts are modeled through Hertzian overlapping.

Increasing the strength and the stiffness of the structural elements the transitions from total to partial collapse and from partial to no collapse are registered. The first collapse mechanism is always triggered by dynamic stress redistribution while impacts make the collapse propagate to the whole system. Three local collapse mechanisms due to impacts have been individuated: hammer effect of falling ceiling slabs onto the underlying ones; drag effect of falling rubble that laterally hits still intact elements; base cutting of the columns due to the lateral pressure exerted by the rubble stacking on the ground.

The future perspective of this work is to optimise the structural response to local damage keeping the cost of the building as low as possible. In order to do this, many simulations must be performed considering different starting damage, geometrical ratios for the frames, structural materials and design strategy. Moreover, geometrical and mechanical disorder, shear walls, bracing and plastic behaviour must be included. Another target of the current research is to increase the size of the studied systems to simulate the progressive collapse of large buildings. This goal can be achieved improving the adopted algorithms (also through parallelisation) and increasing the computational effort.