No abstract

Data acquisition and processing are areas of research in fault diagnosis in rotating machinery, where the rotor is a fundamental component that benefits from dynamic analysis. Several intelligent algorithms have been used to optimize investigations of this nature. However, the Jaya algorithm has only been applied in a few instances. In this study, measurements of the amplitude of vibration in the radial direction in a gas microturbine were analyzed using different rotational frequency and temperature levels. A response surface model was generated using a polynomial tuned by the Jaya metaheuristic algorithm applied to the averages of the measurements, and another on the whole sample, to determine the optimal operating conditions and the effects that temperature produces on vibrations. Several tests with different orders of the polynomial were carried out. The fifth-order polynomial performed better in terms of MSE. The response surfaces were presented fitting the measured points. The roots of the MSE, as a percentage, for the 8-point and 80-point fittings were 3.12% and 10.69%, respectively. The best operating conditions were found at low and high rotational frequencies and at a temperature of 300 ∘C. High temperature conditions produced more variability in the measurements and caused the minimum value of the vibration amplitude to change in terms of rotational frequency. Where it is feasible to undertake experiments with minimal variations, the model that uses only the averages can be used. Future work will examine the use of different error functions which cannot be conveniently implemented in a common second-order model. The proposed method does not require in-depth mathematical analysis or high computational capabilities.

Abstract:Rotating systems components such as rotors, have dynamic characteristics that are of great importance to understand because they may cause failure of turbomachinery. Therefore, it is required to study a dynamic model to predict some vibration characteristics, in this case, the natural frequencies and mode shapes (both of free vibration) of a centrifugal compressor shaft. The peculiarity of the dynamic model proposed is that using frequency and displacements values obtained experimentally, it is possible to calculate the mass and stiffness distribution of the shaft, and then use these values to estimate the theoretical modal parameters. The natural frequencies and mode shapes of the shaft were obtained with experimental modal analysis by using the impact test. The results predicted by the model are in good agreement with the experimental test. The model is also flexible with other geometries and has a great time and computing performance, which can be evaluated with respect to other commercial software in the future.

This work developed the modeling and supervisory control for gas turbine. A CTPN (continuous timed Petri Net) model of a gas turbine, using a first linear order approximation for every state of the Brayton cycle is obtained. The Brayton cycle rules the functioning of a gas turbine, and it is composed by four states: compression, combustion, expansion and cooling. The principle of the gas turbine is developed by the Brayton cycle, a thermodynamic process which intervenes in the gas turbine components. The steady-state behavior of the gas turbine has been widely investigated in engineering area. Moreover, the dynamic behavior has been studied using non-linear models of its components, leading to complicated mathematical representations. The methodology of the current work begins with a simplification of the dynamical relations in every state (excepting the cooling phase) of the Brayton cycle. Temperature and pressure are modeled as first order linear systems, therefore, every system is translated into a CTPN. Furthermore, to guarantee a safety operation, an SC (supervisory controller) is designed to ensure the combustion chamber temperature is lower than 1,000 °C. Although the model presented is extremely simplified, it will be used as a starting point to develop more complex models.

The elements of rotors have inherent characteristics as geometry and material composition, which causes natural vibrations at frequencies that, due to the rotor unbalance, may coincide with the harmonics of the shaft speed, increasing stress and the probability of fractures even in transient conditions. Therefore, in this work, a theoretical-experimental hybrid method for calculating the natural frequencies and the mode shapes, at rest and non-supported conditions, of a turbocharger rotor is proposed. Firstly, a discrete model of low number of degrees of freedom is considered, and from an oversampled modal approach (OSMA) based on the axial oversampling, sectioning and coupling of the rotor, it is possible to use the oversampled mode shapes to increase the degrees of freedom of the system without major complications in the model. This spatial oversampling criterion is based on the Nyquist-Shannon theorem, and it is used to reduce the error in the estimates of the natural frequencies and to get a first approximation of the mode shapes. The natural frequencies were estimated by the transfer matrix method (TMM) and finite element method (FEM) in order to compare the proposed model results with well-founded numerical methods.