A generic element-based method for joint identification

Abstract Modelling complex structural components such as joints is problematic for purely analytical techniques. An alternative technique is to attempt to use experimental data to update or improve finite element models. Model updating techniques may not be suitable for the correction of very complicated structural components, since there will be regions of error throughout the structure, and not merely at the joints. In addition, the assumed finite element (FE) construction of the joint may well be too simple, or even incorrect. This paper examines a technique that calculates the properties of structural joints by minimising the difference between substructure frequency response functions (FRFs) and assembly FRFs. It is shown that conventional joint identification techniques that rely on direct estimation of individual joint model terms do not maintain the connectivity of the model. An optimisation technique is used in this paper to force the joint model to maintain the correct connectivity by updating structural parameters and generic element parameters. These techniques are applied to simulated case studies, where they exhibit significant success. Finally, a true experimental study is undertaken which demonstrates the promise of the proposed method.