DOI QR Code
Framed steel structures frequently have bolted connections to ensure semi-rigid joints that have a significant amount of energy dissipation incorporated in them, to avoid failure of connections and members under cyclic loading, such as wind and earthquake loading. One problem is that due to cyclic action, these bolts may loosen, and compromise subsequent behavior of the structure. A vibration-based health monitoring technique for quantification of the level of loosening of bolted joints in a steel plane frame structure is presented here, to provide a basis for evaluation of the structure state. A numerical model of a plane frame is considered with rotational springs representing semi-rigid joints. A fixity factor is defined in terms of rotational spring stiffness and is considered as a measure of level of loosening of bolts with zero representing fully loose and one representing fully tight condition. Experimental strain time histories are collected and transformed into frequency domain using Fourier transform. A shape co-relation is defined using frequency data obtained from the damaged and the undamaged structures. Using the frequency spectra and shape correlation, an objective function (OF) is developed and minimized by the particle swarm optimization to estimate the fixity factor. It is found that the technique estimates higher value of reduction of the fixity factor in the damaged location, but shows some considerable value at the other springs also. Therefore, the technique is improved using heuristics by identifying probable damage locations prior to applying model updation, in order to estimate the damage severity more accurately. Considering fixity factors at the identified locations as variables, model updating is done for estimation of fixity factors. The improved results clearly indicate actual damage locations and fixity factors for different levels of bolt loosening, and is found satisfactory for possible future application of the technique to multistory framed structures.