Quantifying effects of system-wide uncertainties (i.e., affecting structural, piezoelectric, and/or electrical components) in the analysis and design of piezoelectric vibration energy harvesters have recently been emphasized. The present investigation proposes first a general methodology to model these uncertainties within a finite element model of the harvester obtained from an existing finite element software. Needed from this software are the matrices relating to the structural properties (mass, stiffness), the piezoelectric capacitance matrix as well as the structural-piezoelectric coupling terms of the mean harvester. The thermal analogy linking piezoelectric and temperature effects is also extended to permit the use of finite element software that do not have piezoelectric elements but include thermal effects on structures. The approach is applied to a beam energy harvester. Both weak and strong coupling configurations are considered, and various scenarios of load resistance tuning are discussed, i.e., based on the mean model, for each harvester sample, or based on the entire set of harvesters. The uncertainty is shown to have significant effects in all cases even at a relatively low level, and these effects are dominated by the uncertainty on the structure versus the one on the piezoelectric component. The strongly coupled configuration is shown to be better as it is less sensitive to the uncertainty and its variability in power output can be significantly reduced by the adaptive optimization, and the harvested power can even be boosted if the target excitation frequency falls into the power saturation band of the system.

Component-centric reduced order models (ROMs) have recently been developed in the context of linear structural dynamics. They lead to an accurate prediction of the response of a part of structure (referred to as the β component) while not requiring a similar accuracy in the rest of the structure (referred to as the α component). The advantage of these ROMs over standard modal models is a significantly reduced number of generalized coordinates for structures with groups of close natural frequencies. This reduction is a very desirable feature for nonlinear geometric ROMs, and thus, the focus of the present investigation is on the formulation and validation of component-centric ROMs in the nonlinear geometric setting. The reduction in the number of generalized coordinates is achieved by rotating close frequency modes to achieve unobservable modes in the β component. In the linear case, these modes then completely disappear from the formulation owing to their orthogonality with the rest of the basis. In the nonlinear case, however, the generalized coordinates of these modes are still present in the nonlinear stiffness terms of the observable modes. A closure-type algorithm is then proposed to finally eliminate the unobserved generalized coordinates. This approach, its accuracy and computational savings, is demonstrated first on a simple beam model and then more completely on the 9-bay panel model considered in the linear investigation.

Component-centric reduced order models (ROMs) are introduced here as small-size ROMs providing an accurate prediction of the linear response of part of a structure (the β component) without focusing on the rest of it (the α component). Craig–Bampton (CB) substructuring methods are first considered. In one method, the β component response is modeled with its fixed interface modes while the other adopts singular value eigenvectors of the β component deflections of the linear modes of the entire structure. The deflections in the α component induced by harmonic motions of these β component modes are processed by a proper orthogonal decomposition (POD) to model the α component response. A third approach starts from the linear modes of the entire structure which are dominant in the β component response. Then, the contributions of other modes in this part of the structure are approximated in terms of those of the dominant modes with close natural frequencies and similar mode shapes in the β component, i.e., these nondominant modal contributions are “lumped” onto dominant ones. This lumping permits to increase the accuracy in the β component at a fixed number of modes. The three approaches are assessed on a structural finite element model of a nine-bay panel with the modal lumping-based method yielding the most “compact” ROMs. Finally, good robustness of the ROM to changes in the β component properties (e.g., for design optimization) is demonstrated and a similar sensitivity analysis is carried out with respect to the loading under which the ROM is constructed.

This paper focuses on the formulation and validation of a novel perturbation method for the prediction of the forced response of mistuned bladed disks. At the contrary of most previous methods, this approach leads to a convergent series representation over the entire range of blade–disk coupling levels for small mistuning. The dominant parameter affecting the magnitude of the largest mistuning for which convergence occurs is shown to be the system damping with a weaker effect of the blade–disk coupling. Examples of application on a single degree-of-freedom per blade model and the reduced order model of a blisk demonstrate the potential of this novel approach. Finally, the applicability of this technique for the optimization of intentional mistuning pattern is shown.

This paper focuses on the development of nonlinear reduced order modeling techniques for the prediction of the response of complex structures exhibiting “large” deformations, i.e., a geometrically nonlinear behavior, which are nonintrusive, i.e., the structure is originally modeled within a commercial finite element code. The present investigation builds on a general methodology successfully validated in recent years on simpler beam and plate structures by: (i) developing a novel identification strategy of the reduced order model parameters that enables the consideration of the large number of modes (>50 say) that would be needed for complex structures, and (ii) extending a step-by-step strategy for the selection of the basis functions used to represent accurately the displacement field. The above novel developments are successfully validated on the nonlinear static response of a nine-bay panel structure modeled with 96,000 degrees of freedom within Nastran.

This paper focuses on the optimization of intentional mistuning patterns for the reduction of the sensitivity of the forced response of bladed disks to random mistuning. Intentional mistuning is achieved here by using two different blade types (denoted as A and B) around the disk. It is thus desired to find the arrangement of these A and B blades (A/B pattern) that leads to the smallest 99th percentile of the amplitude of blade response in the presence of random mistuning. It is first demonstrated that there usually is a large number of local minima and further that the cost of a function evaluation is high. Accordingly, two novel, dedicated optimization algorithms are formulated and validated to address this specific problem. Both algorithms proceed in a two-step fashion. The first step, which consists of an optimization in a reduced space, leads to a series of good initial guesses. A local search from these initial guesses forms the second step of the methods. A detailed validation effort of this new procedure was next achieved on a single-degree-of-freedom-per-blade model, a reduced order model of a blisk, and that of an impeller considered in an earlier study. In all validation cases, the two novel algorithms were found to converge to the global optimum or very close to it at a small computational cost. Finally, the results of these optimization efforts further demonstrate the value of intentional mistuning to increase the robustness of bladed disks to random mistuning.

This paper focuses on extending an earlier investigation on the systematic and rational consideration of uncertainty in reduced order models of rotordynamics systems. The current effort concentrates on the consistent introduction of uncertainty in mass properties on the modal mass and gyroscopic matrices and on the unbalance force vector. The uncertainty in mass is separated into uncertainty that maintains the rotor symmetry and the one which disrupts it. Both types of uncertainties lead to variations in the system modal matrices but only the latter induces an unbalance. Accordingly, the approach permits the selection of separate levels on the uncertainty on the system properties (e.g., natural frequencies) and on the unbalance. It was first found that the unbalanced response is increased by considering the uncertainty in the rotor modal mass matrices. It was next noted that the approach presented not only permits the analysis of uncertain rotors but it also provides a computational framework for the assessment of various balancing strategies. To demonstrate this unique feature, a numerical experiment was conducted in which a population of rotors were balanced at low speed and their responses were predicted at their first critical speed. These response predictions were carried with the uncertainty in the system modal mass matrices but with or without the balancing weights effects on these matrices. It was found that the balancing at low speed may, in fact, lead to an increase in both the mean and 95th percentile of the response at critical speed.