Introduction
There has been a rising interest among engineers to conduct non-linear time history (NLTH) analysis as part of the seismic design procedure for bridges. A good understanding of member parameters such as strength, stiffness and damping are crucial for accurate results from NLTH analysis. Strength and stiffness in dynamic analyses are well understood but the viscous damping model that is mostly being used in commercial analysis programs, namely initial stiffness proportional Rayleigh damping, is questionable. This model prescribes a constant damping matrix proportional to the initial stiffness matrix of a structure. Theoretically, this is inaccurate because of three main reasons.
Elastic viscous damping that accounts for material damping in the elastic range is not required to continue the same function in the inelastic range where hysteretic damping exists.
Since the stiffness of the system changes in the inelastic range, an argument can be made against the use of a constant damping proportional to initial stiffness.
Elastic viscous damping also accounts for energy dissipation by the foundation (mainly in bridges). Since the demand going into the foundation remains constant after yielding, using the same damping in the inelastic range is erroneous.
Albeit, the above can be stated in theory, if the sensitivity of response prediction to the variation in damping models is minor, the simplest model would serve the purpose. Sensitivity of displacement predictions of SDOF systems to choice of initial stiffness and tangent stiffness proportional damping models in NLTH analysis have been studied in the past. It was found that the results vary considerably, with the tangent stiffness proportional model better representing actual tests. This project makes an effort towards understanding the sensitivity of response predictions of MDOF systems to the choice of four different damping models in NLTH analysis.
Research Report
For more details on this project, please read this report.