Structural damage identification: influence of model incompleteness and errors

This paper presents a study of the influence of model incompleteness and errors in a structural damage identification technique based on modal data, leading to a set of linear equations. To deal with the incompleteness three methods are used: (1) dynamic and (2) static expansions of measured degrees of freedom and (3) replacement of unmeasured degrees of freedom by their undamaged counterparts. These methods are applied on damage identification of a laminated rectangular plate. The numerical tests show that the best accuracy is obtained with the dynamic expansion, followed by the static expansion and finally, the replacement of unmeasured degrees of freedom. It is also demonstrated that for small damage the errors are the main influence, whereas for large damage the incompleteness becomes the most important factor in the results. The identification of small and large damages is performed by deleting non-reliable equations, i.e., equations that contain large errors.