Force and response estimation on bottom-founded structures prone to ice-induced vibrations

Abstract

Various types of bottom-founded structures, including lighthouses, quay structures, mono-pod platforms, multi-legged platforms, caisson-retained islands and bridges, are located in ice-infested waters. Level ice can interact with bottom-founded structures in various manners, and over fifty years of extensive measurement campaigns has brought attention to ice-induced vibrations. This phenomenon is caused by repeated ice crushing failures across the ice-structure interface and may entail violent vibrations of the structure, thereby potentially harming the structural integrity, secondary installations and operational safety. Such ice-induced vibrations are commonly divided into three regimes: 1) Intermittent crushing 2) Frequency lock-in 3) Continuous brittle crushing in which the ice velocity increase from regime 1 to regime 3. The ice conditions leading to each of the three regimes are not yet fully understood. Therefore, measurement campaigns both in the field and in the laboratory must address these regimes, wherein two of the major ingredients are the ice force and the structural response. A laboratory-scale ice-induced vibration measurement campaign was conducted at the Hamburg Ship Model Basin during August-September 2011, from which data were obtained for this thesis. Data measured at the Nordströmsgrund lighthouse in Sweden during the winter of 2003 and structural information on the Hanko-1 channel marker in Finland constitute the full-scale basis in this thesis. The ice forces present during ice-induced vibrations are traditionally measured by load panels or inverse techniques. Load panels are expensive; thus, inverse techniques are favorable. This thesis assessed a deterministic-stochastic framework to identify both the ice forces and responses at both the model scale and full scale. All of the considered data were limited to scenarios of ice-induced vibrations, and the considered ice conditions were primarily level ice. The framework as it is applied in this thesis consists of a joint input-state estimation algorithm, a model of the structure and a set of response measurements. Both full-order finite element models and modally reduced order models were used in this thesis. Using the laboratory measurements, the force and response identification was performed by employing two different full-order finite element models. One model was entirely based on the blueprints of the structure. The other model was tuned to more accurately reproduce the measured first natural frequency. The results were presented for two different regimes of ice-induced vibrations: the intermittent crushing regime and the continuous brittle crushing regime. The accuracy of the identified forces using the joint input-state estimation algorithm was assessed by comparing the forces with those obtained by a frequency-domain deconvolution method based on experimentally obtained frequency response functions. The results demonstrated the successful identification of the level-ice forces for both the intermittent and continuous brittle crushing regimes even when significant modeling errors were present. The responses (displacements) identified in conjunction with the forces were also compared to those measured during the experiment. Here, the estimated response was found to be sensitive to the modeling errors in the blueprint model. Simple tuning of the model, however, enabled high-accuracy response estimation. The joint input-state estimation algorithm was further used as a means to analyze the laboratory data, from which the global structural response was simultaneously identified with the forces. Novel insights into ice-induced vibration phenomena were obtained by comparing, on different time scales, measured and estimated response quantities and forces/pressures. First, the identified forces, ice velocities and time-frequency maps of the measured responses were presented for a series of ice-induced vibration tests. It was shown that the ice forces excited more than one mode of the structure and that the transition ice velocity at which the vibrations shifted from the first mode to the second mode increased with decreased foundation stiffness and superstructure mass. Second, a detailed analysis of the interaction between the structure and the ice edge was performed on a smaller time scale by comparing the locally measured pressures at the ice-structure interface to the identified structural responses and forces. It was shown that structural vibrations at a frequency that is higher than the dominant vibration frequency caused cyclic loading of the ice edge during intermittent crushing. These vibrations led to an increasing loading rate prior to ice failure. During an event that showed the tendencies of frequency lock-in vibrations, the structural response was dominated by a single vibration frequency. At full scale, a comparison between the measured and identified dynamic ice forces acting on the Nordströmsgrund lighthouse is presented. The dynamic ice forces were identified from the measured responses using the joint input-state estimation algorithm in conjunction with a reduced-order finite element model. A convincing agreement between the measured and identified forces was found. The algorithm was further used to estimate the response of the structure at unmeasured locations, including the iceaction point. The structural velocity amplitudes when the structure was subject to frequency lock-in vibrations were occasionally higher than the ice velocity and within the range of observations for other structures. A measurement campaign at the Hanko-1 channel marker in the Gulf of Finland is planned to monitor the forces leading to ice-induced vibrations via force identification. The ice forces are to be identified using the joint input-state estimation algorithm in conjunction with a modally reduced order model. Recently developed guidelines were used to determine the optimal response measurement types and locations that ensure the identifiability of the dynamic ice forces from only a limited number of sensors and a selection of vibration modes.