Data-Driven Model for the Dynamic Characteristics of O-Rings for Gas Bearing Supported Rotors

The measurement results of various nitrile butadiene rubber (NBR) O-Ring sizes are presented, and reduced-order models are developed in order to predict the stiffness and damping coefficient as a function of O-Ring geometry, Shore hardness, squeeze, and excitation frequency. The results show that the curvature ratio d/D needs to be considered in the reduced-order models. The assessment of the model suggests a maximum deviation of 30% in predicted stiffness compared to the measurement data. However, taking into account the typical Shore hardness tolerance given by O-Ring manufacturers and other measurement uncertainties, the proposed model enables the prediction of various O-Rings with a good accuracy in the frequency range of 1.5–3.75 kHz, which corresponds to typical gas bearing supported rotor applications.

Download Full-text

A New Mistuning Identification Method Based on the Subset of Nominal System Modes Method

Journal of Engineering for Gas Turbines and Power ◽

10.1115/1.4045517 ◽

2020 ◽

Vol 142 (2) ◽

Author(s):

Christian U. Waldherr ◽

Patrick Buchwald ◽

Damian M. Vogt

Keyword(s):

Finite Element ◽

Input Data ◽

Measurement Data ◽

Computational Effort ◽

Mode Shapes ◽

Reduced Order Models ◽

Identification Procedure ◽

Reduced Order ◽

Nominal System ◽

Identification Method

Abstract The mistuning problem of quasi-periodic structures has been the subject of numerous scientific investigations for more than 50 years. Researchers developed reduced-order models to reduce the computational costs of mistuning investigations including finite element models. One question which has also high practical relevance is the identification of mistuning based on modal properties. In this work, a new identification method based on the subset of nominal system modes method (SNM) is presented. Different to existing identification methods where usually the blade stiffness of each sector is scaled by a scalar value, N identification parameters are used to adapt the modal blade stiffness of each sector. The input data for the identification procedure consist solely of the mistuned natural frequencies of the investigated mode family as well as of the corresponding mistuned mode shapes in the form of one degree-of-freedom per sector. The reduction basis consists of the tuned mode shapes of the investigated mode family. Furthermore, the proposed identification method allows for the inclusion of centrifugal effects like stress stiffening and spin softening without additional computational effort. From this point of view, the presented method is also appropriate to handle centrifugal effects in reduced-order models using a minimum set of input data compared to existing methods. The power of the new identification method is demonstrated on the example of an axial compressor blisk. Finite element calculations including geometrical mistuning provide the database for the identification procedure. The correct functioning of the identification method including measurement noise is also validated to show the applicability to a case of application where real measurement data are available.

Download Full-text

A New Mistuning Identification Method Based on the Subset of Nominal System Modes Method

Volume 7B: Structures and Dynamics ◽

10.1115/gt2019-91702 ◽

2019 ◽

Author(s):

Christian U. Waldherr ◽

Patrick Buchwald ◽

Damian M. Vogt

Keyword(s):

Finite Element ◽

Input Data ◽

Measurement Data ◽

Computational Effort ◽

Mode Shapes ◽

Reduced Order Models ◽

Identification Procedure ◽

Reduced Order ◽

Nominal System ◽

Identification Method

Abstract The mistuning problem of quasi periodic structures is subject of numerous scientific investigations for more than 50 years. Researchers developed reduced order models to reduce the computational costs of mistuning investigations including finite element models. One question which has also high practical relevance is the identification of mistuning based on modal properties. In the present work, a new identification method based on the Subset of Nominal System Modes method (SNM) is presented. The input data for the identification procedure consists solely of the mistuned natural frequencies of the investigated mode family as well as of the corresponding mistuned mode shapes in the form of one degree of freedom per sector. The reduction basis consists of the tuned mode shapes of the investigated mode family. Furthermore, the proposed identification method allows for the inclusion of centrifugal effects like stress stiffening and spin softening without additional computational effort. From this point of view the presented method is also appropriate to handle centrifugal effects in reduced order models using a minimum set of input data compared to existing methods. The powerfulness of the new identification method is demonstrated on the example of an axial compressor blisk. Finite element calculations including geometrical mistuning provide the data base for the identification procedure. The correct functioning of the identification method including measurement noise is also validated to show the applicability to a case of application where real measurement data is available.

Download Full-text