[Back]

A. Kazemi Amiri:
"Inverse reconstruction of wind load and stochastic response analysis from sparse long-term response measurements";
Supervisor, Reviewer: C. Bucher, R. Höffer, C. Adam; Vienna Doctoral Programme on Water Resource Systems, 2016; oral examination: 2016-10-10.

Wind can affect a wide range of structures including ordinary buildings, high-rise buildings and towers, overhead power lines, on/offshore wind turbines, cranes and industrial chimneys etc. In this sense the wind loading accounts for the destructive effects on the structures, which - depending on the particular case - can be due to wind overloads in storm event (e.g. hurricanes, typhoons), aeroelastic stability issues, architectural damages due to sudden change in wind pressure gradients or cumulative fatigue damage in structural elements. As a result wind loading of the structures has received substantial research works in the past decades. Due to this reason, the main attention in this dissertation was drawn to the wind-induced vibration of structures among other excitation sources. The primary goal of the dissertation is inverse identification of the wind load, which is the source of wind-induced vibrations. By -inversely- it is pointed out that wind load cannot be easily measured directly and it is recovered from its effect on the structure, i.e. from the structural response. To this end new formulations to derive the impulse response matrix is provided, which is then used in the problem of load identification. The ill-conditioning of the impulse response matrix made it necessary to deploy a regularization scheme to recover the applied force from noise polluted measured response. The Tikhonov regularized solution in conjunction with generalized cross validation (GCV) and L-curve method were used to solve the inverse problem. The identification procedure was implemented for a simple simulation example as well as its corresponding experimental laboratory case. It is shown that the accuracy of experimentally identified load depends on the sensitivity of measurement instruments over the different frequency range. In the next step, a procedure for inverse wind load reconstruction is presented, which is applicable to multiple degrees of freedom system and is especially suitable for practical purposes. For the sake of higher accuracy and computational efficiency the load identification is performed in the modal subspace. In this way just the modal parameters of a system namely eigenfrequencies and -vectors as well as the damping ratios should be known. It is investigated, which response type is more appropriate for the proposed wind load reconstruction procedure. The results of problem simulations for a real structure demonstrate that the modal wind loads can be successfully identified more accurately from displacement than acceleration response even at relatively high noise levels. Afterwards the field application of the introduced procedure for the wind load identification was carried out. The structure under measurement is a 9.1 m (30 ft) tall guyed mast. The modal wind loads are identified in modal subspace of the mast for several single degree of freedom systems, whose characteristic parameters are obtained by an operational modal analysis procedure. The experimentally reconstructed modal loads were verified by inspecting the analogy between field and simulation results. Since Vienna doctoral program on water resource systems (DK) is a multidisciplinary program, collaborative research works between and within research clusters of the DK is one of the main focuses of the doctoral program. The author-s contribution to the collaborative research work consists of two parts. The first part pursues the second goal of the dissertation, which is stochastic response analysis of a structure assisted by mean wind speed data, when just the structure discontinuous/sparse response data at least within one year is available. The outcome of such study is remarkably helpful to the structural vibration control under wind excitations. The wind speed data was provided by the weather station, belonging to the Hydrological Open Air Laboratory (HOAL) of the DK. Thereby histogram and accordingly the mean wind speed probability distribution function of different blowing directions were obtained. Every triggered structural data was tagged by its associated mean wind speed data. The structural acceleration of the mast was measured according to an 18-hour automatic trigger. The ten-minutes acceleration data was recorded after each triggering. Then the mathematical relationship between mean wind speed data and response standard deviation, displacement response threshold passage counts and moments of area of the stress power spectral density was established. The author-s second collaboration may not directly contribute to attain the objectives of the dissertation, but the applied theories are pretty relevant to the content of the dissertation. In this collaboration, the methodical developments of a new model order reduction (MOR) strategy based on the proper orthogonal decomposition (POD) method, which applies to the nonlinear dynamic problems, are presented. An academic example structure with bilinear elastoplastic material behavior as well as a realistic hospital complex with single frictional base isolators were used to assess the introduced method. The results demonstrate accurate approximations of the physical (full) responses by means of this new MOR strategy if the probable behavior of the structure has already been captured in the POD snapshots. The dissertation tried to improve the efficiency of the impulse response matrices of structural systems and developed a practical procedure for inverse reconstruction of wind loads on the structure, only based on the data that can be achieved via measurement in reality. The dissertation provided an effective method for long-term stochastic response analysis of structures under wind excitation, while continuous response measurement is no longer needed. This method can also be deployed in numerical simulations to achieve more realistic long-term response analysis of structures under wind.

wind load; stochastic response analysis