Evaluation of analytical and experimental methods to predict constrained layer damping behavior
Constrained layer damping (CLD) for three layer beams with viscoelastic cores and aluminum or graphite/epoxy composite laminate outer layers was investigated for model comparison, sensitivity to design parameters, and evaluation of experimental loss factor estimation techniques. Model comparison for damping estimation and resonant frequency prediction was performed between finite element analysis (FEA) Ross, Kerwin, and Ungar theory (RKU), developed moment predicted loss factor, and experimental results. Investigated design parameters include; treatment application length and placement, relative thickness of core and constraining layers to base layer, core loss factor, and boundary conditions (free/free, fixed/fixed, and cantilever). Experimental damping estimation techniques evaluated include; frequency response function (FRF) based methods of component analysis, circle-fit method a curve-fit algorithm developed by Han  and the time domain of log decrement.
Model comparison of finite elements to experimental results showed good trend prediction correlation. Only fair prediction of absolute loss factors was achieved, possibly due to the difficulty in characterization of viscoelastic properties. Design parameters analysis showed that treatment application length and placement were critical to effective added damping. In one case, for the same amount of damping material, the effective added damping of a well designed application was seven times greater than that of a poorly considered one. The effectiveness of treatment on a region appears to be strongly related to the magnitude on the moment acting on that region. Parameter analysis also showed that although a symmetric beam realizes the highest damping, in most cases near optimal damping can be obtained with constraining layer one half as thick as the base layer. Experimental methods for damping estimation showed the simple FRF component analysis to be consistent with the other methods (experimental and FEA) and the most computationally efficient.