Uncertainty evaluation of semi-active load redistribution in a mechanical load-bearing structure

Load-bearing structures in mechanical engineering applications typically face the challenge of withstanding and transmitting external loads. In most cases, the load path through the load-bearing structure is predetermined by the design. However, if parts of the load-bearing structure become weak or suffer damage, e.g. due to deterioration or overload, the load capacity becomes uncertain. In this thesis, the semi-active load redistribution to bypass a portion of the loading away from damaged parts of the structure is used in order to prevent the structure from failure or malfunction. So far, studies on semi-active or active measures to adapt or manipulate the dynamic behavior of a structure have primarily investigated damping or vibration control and not load redistribution. The proposed semi-active load redistribution provides a technological possibility to influence the load path during operation via augmenting already existing parts of the load-bearing structure with actuators. Furthermore, for accurate numerical predictions of the load redistribution capability, an adequate mathematical model is needed. Therefore, the accuracy of the load-bearing structure’s mathematical model predictions is evaluated and increased methodologically by model parameter uncertainty quantification and reduction. The structure to numerically and experimentally investigate load Redistribution in this thesis is based on a load-bearing structure developed within the SFB 805 and consists of a translational moving mass connected to a beam by a spring-damper system and two newly developed semi-active augmented guidance elements for load redistribution. The beam is supported at its ends by two supports. The stiffness characteristic of the supports can be adjusted to simulate structural damage. The structural damage, in turn, causes misalignment of the beam, which is defined as malfunction. A mathematical model of the load-bearing structure is derived for numerical investigations of the load redistribution capability and for controller design. A BAYESIAN inference based calibration procedure is applied to reduce and simultaneously quantify the model parameter uncertainty. Thus, the model is adjusted to the present conditions and the model prediction accuracy is increased. Clipped-optimal LQR and PID controllers are introduced for the semi-active load redistribution and designed based on the calibrated model. With the presented procedure, the model prediction variation due to Parameter uncertainty is reduced by up to 85%. Comparing the passive and semi-active load-bearing structure, the malfunction is reduced by up to 53% numerically and by up to 51% experimentally. The evaluation of the load paths shows that a redistribution of the load between the two supports is achieved by means of the semi-active guidance elements. The results of this thesis contribute to the methodological parameter uncertainty quantification and reduction as well as the technological application of semi-active load redistribution.