The paper presents a general method for modal decomposition of time series of structural vibration response. The method is applied to measurements of vortex induced vibrations of deep water drilling risers in operation. The modal decomposition method is based on a subspace system identification algorithm that is driven by the measured riser and rig response data only. Thus no a priori FEM model of the riser system is required for determination of the mode shape matrix that is used in the modal decomposition of the measured response time series. A combined deterministic-stochastic model of the riser-rig dynamic system is identified from response measured at fixed positions along the riser and from measurements of the rig motions. The rig motions are considered as a deterministic input to the dynamic system. The linear riser dynamics caused by the rig motions can therefore be separated from the riser vibrations caused by other unmeasured (or in this context by definition: stochastic) excitation sources, such as e.g. vortex shedding. It is well known that hydrodynamic damping is significant for deepwater risers. Such damping is not classical in the sense of Caughey. The dynamic system model must therefore allow for general damping properties. This implies complex eigenvectors of the associated damped eigenvalue problem, and complex modal coordinates of the decoupled system. It will be shown how the elements of the complex eigenvectors can be interpreted in terms of magnitudes and phase angles of the corresponding mode shapes. The complex modal coordinate time series are interpreted as modal amplitude and modal phase angle. The modal decomposition method is illustrated by application to a few sets of measured response time series from a deep-water drilling riser.
The present trend in offshore petroleum exploration is towards deeper waters.
