2004 · 20 citations · 5 references
EngineeringCivil EngineeringStructural AnalysisStructural DynamicsStructural EngineeringModal AnalysisModal TheoryMode ShapesStructural MechanicsStructural Response AnalysisVibration AnalysisInput-output Modal AnalysisModal Parameter EstimationModal ScalingModal TestingMode Shape EstimationMode Shape ModelingMode Shape ScalingScaling Factor Estimation
It is well known, that when using natural input modal analysis, the loads are not known, and thus, the mode scaling factor that relates the magnitude of the loading to the magnitude of the response cannot be estimated. However It has been pointed out by several theoretical papers that mode shapes can be scaled by performing several natural input modal analysis tests with different mass changes, observe the frequency shift introduced by the mass changes and then follow an estimation scheme that allows the user to estimate the scaling factor modeby-mode, i.e. only information of the particular mode of interest is used to obtain the scaling factor for that mode. The procedure is of high practical interest also in mechanical engineering since it is well known that the traditionally estimated scaling factor is often suffering from large uncertainties. In this paper it is shown how the mass change technique can be used on a 1⁄4 scale model of a 4-storey building. The uncertainties on the estimated scaling factors are illustrated by repeating the estimation using different mass changes. INTRODUCTION Output-only modal testing and analysis – or as it may be should be called: Natural input modal testing and analysis is becoming more and more popular due to the clear advantages of the technology: The testing is easier, the technology is applicable to a wider a range of the structures and it has a wider range of potential applications since the actual responses are stored and can be used for instance fatigue analysis and vibration level estimation. Also the technology gives better and more reliable results in cases where the actual loading conditions and operating conditions are important for the structural response. However the technology is still in the early stages of development and many problems still remain unsolved or partly unsolved. One of the important remaining problems is the problem of mode shape scaling. If the identified modal model is going to be used for structural response simulation or for structural modification, then the scaling factors of the mode shapes must be known. Also in health monitoring applications and in cases where damage is to be identified, the scaling factors might be useful. Recently some suggestion has been given in the literature for solving this problem. One solution has been suggested by Bernal and Gunes [1] based on the assumption that partition of the inverse of the mass matrix associated with the measured coordinates is diagonal. However, the approach gives exact answers only when there is a full set of modes, and robustness for a truncated modal space has not been demonstrated. Recently Parloo et al. [2] have published a new approach based on a more extensive testing procedure that involves repeated testing where mass changes are introduced in the points of the structure where the mode shape is known. This approach seems more appealing, since to scale a certain mode, only that particular mode has to be known. Brincker et al. [3], [4] have published an improved procedure following the ideas of Parloo et al. and also performed an uncertainty analysis that clearly indicates a lower uncertainty on the mode shape when following this estimation procedure for the scaling factor than what is know to be the case using traditional modal analysis. In Brincker et al. [3], [4] it is shown that if one arbitrary mode shape φ estimated by natural input modal analysis is scaled to unity, i.e. so that 1 = φ φT and the corresponding mode shape ψ scaled as usual so that 1 = Mψ ψT , then the scaling factor α relating the two mode shape estimates α = ψ φ can be found by introducing a mass change and obtaining the estimate as 2 2 1 2 2 2 ω ω α ω − = T φ ∆Mφ (1) where ∆M is the mass change matrix and 1 ω , 2 ω are the natural frequency of the considered mode before and after application of the mass change. If the mass change matrix is proportional to the initial mass matrix, then the mode shapes will not change, and equation (1) is then an exact linear elastic solution for the relationship between mass changes, frequency shift and mode shape scaling factor. In the case where the mass change matrix is not proportional to the initial mass matrix, equation (1) becomes approximate. Assuming that the uncertainty α σ on α is controlled by the uncertainty ω σ ∆ on the frequency shift ω ∆ and by the uncertainty φ σ on the mode shape coordinates, then, if the mass change matrix can be written D ∆M m ∆ = , where the matrix D is a diagonal matrix with only unity values and zeroes in the diagonal, then the uncertainty on the scaling factor can be estimated from
5