Seismic analysis of knee elements for steel frames
The 1994 Northridge and 1995 Kobe earthquakes, which were moderate in seismological terms, showed that many buildings were subjected to demolition or very expensive repairs because of severe damage in principle members, mainly in the column-beam connections. As a result, the development of dissipative systems was encouraged, which limit the damage parts to easily replaceable elements, in case of moderate earthquakes. One such system is the knee braced frame. Knee braced frames are a modified form of cross bracing in which the brace is cut short and connected to the mid point of a knee element spanning between the adjacent beam and column. The key component is the knee element, which controls both the initial elastic stiffness of the frame, and the onset of yield and subsequent energy dissipation. The knee elements are required to ensure energy absorption through repeated large deformations without suffering collapse or instability. This thesis describes the development of different knee element designs and their performance assessments. It is shown that the dissipative mechanism of the web yielding in shear is advantageous because it is independent of the moment distribution and it does not affect the connections and extends the dissipative zones to all its lengths. Extensive finite element modelling and experimental testing have been undertaken. In the shear yielding mode excellent performance was achieved using standard hot rolled sections, modified by the addition of web stiffeners to prevent localised buckling failure. Weakening of the knee element's webs so that it yields very early in an earthquake has potential benefit, but is shown to be unsafe as it promotes premature failure of the element. A knee element model for non-linear dynamic analysis of an entire building has been developed. Time history analyses showed that knee braced frames with the developed knee element have a large global ductility and an outstanding performance. Results obtained with different pushover analysis methods (Eurocode 8, FEMA-356 and ATC-40) have been compared to those obtained wit the time history analyses. Moreover FEMA-356 method, which includes a more accurate representation of the structure's significant post-yield stiffness, gave the closest agreement with the time history analyses and is recommended for the design of knee braced frames.