Condition monitoring of welding processes have received considerable attention in recent years. The method proposed in this paper provides a novel and a better method for analysis of the weld joint strength, i.e., the adaptive chirplet transform. The presence of the nonlinearities in the various sensor outputs of the monitoring systems of the welding procedure demands a more precise signal processing method for a more accurate analysis of the weld joint strength. The adaptive chirplet method has been used here which produced much better results than the other statistical signal processing methods like the wavelet transform technique due to a better time-frequency resolution of the same. In nonlinear feature extraction, wavelet transform technique was first used to detect the weld joint strength using current as a sensor output during the welding. Then the similar procedure was followed using the adaptive chirplet analysis technique which not only showed better differencing capacity between various signals but also provided better time-frequency resolution for the experimental cases where the wavelet method could not predict the weld joint strength correctly. A thorough laboratory study shows that the diagnostic method proposed in this paper is much more accurate, has high sensitivity with respect to faults, and also has better diagnostic resolution.

No abstract

Most fully analytical treatments of rotor whirl are restricted to very simple rotor geometries, and have limited use while studying real-world rotors. Analysis of more complex rotors is routinely carried out in Finite Element software such as Ansys, but this is less effective at generating fundamental insights to inform the design process. This paper is motivated by the ongoing development of a Supercritical-CO2 turbine rotor, and illustrates a rotor model of intermediate complexity. Such a model, given the relative novelty of the turbine design task stemming from the very high rotational speeds required, will improve confidence in rotordynamic simulations and is expected to help with any future troubleshooting. The approach used is to consider the rotor’s rigidity and inertia to vary along its length as functions of spatial coordinate x, and obtain predictions of whirl speeds through a Lagrangian formulation with assumed modes. This method can accommodate axisymmetric rotors with variation in cross-section or material properties, or both. It is first demonstrated on a simple and analytically tractable rotor geometry. Then, it is applied to a simplified version of the turbine rotor, whose properties’ x-dependence is approximated by curves fit to a small number of datapoints obtained from simpler (non-rotordynamic) simulations.

No abstract