Interface Reduction in Component of Bladed Disks by different methods = Different Interface Reduction Methods for Dynamic Analysis of Bladed Disks

This thesis work has the aim to study a rotor disk of a compressor for aeronautical applications. It's known, that if a rotor disk of a compressor is considered, from a failure point of view, the blades are the most critical component. The main cause derives mostly from vibrations generated from external forces. This phenomenon it's cause of a reduction of the time to failure of the compressors. In this thesis, first some theory about systems of compressor and mistuning is given. This helps to understand the dynamic behaviour of a rotor disk and how it changes when the geometry of the disk is slightly different from the nominal one and cannot be modelled under cyclic symmetry conditions. Deviation of the disk's geometry from the nominal one is known to us as “Mistuning”. More simply, mistuning means that the rotating disk is not perfectly symmetric along the rotation axis. To study the effects of mistuning on the dynamic behaviour of the disk first an approximation of the system is needed. This task is very demanding because it depends on many factors and some of them are difficult to predict. In this work a “Full disk model” geometry has been developed with a two degrees of freedom (DoF's) per sector simplified model. In this model, all the sectors of the rotor wheel have been considered and modelled. After this, a “Sector cyclic model”, made with only one disk sector of three degrees of freedom (DoF's) then reduced to two, has been developed. It's better to specify that one sector comprehends a slice of disk and one blade. For this reason, the number of sectors will be always equal to the number of blades. The "Sector cyclic model" it's simpler than the "Full disk model" because it considers only one sector instead of all of them. In the "Sector cyclic model" the single sector is then repeated through a phase angle in order to obtain the results for the full disk. Dynamic results from the "Full disk model" (the benchmark) and from the "Sector cyclic model", under appropriate model and calculus assumption, have been compared. The aim of the comparison is to check if dynamic analysis performed on a single sector are reliable and gives dynamic response with enough degree of accuracy. Since a model of two degrees of freedom (DoF's) per sector it's too approximative, another system of twenty-four blades disk is developed with the help of the software ANSYS. This model will be the benchmark for the further steps. The goal of this second part is to investigate about the possibility to find a reduced model of the disk that can approximate the study of a rotor wheel without mistuning. In this manner it will be possible to study the full model considering few nodes. These nodes are called “master nodes”. The reduction of the number of nodes has been done on the interfaces between the sectors of the disk. This reduction was done with the help of transformation matrix obtained with different methods. These methods are respectively; SEREP (System equivalent expansion reduction process), Four-noded rectangular element reduction, Eight-noded rectangular element reduction and Nine-noded rectangular element reduction. In the conclusions it's shown which method better approximates the results obtained from the model developed on ANSYS. The results of this work could be than used to study a rotor wheel in presence of mistuning.