Session: 26-02 Applications using Probabilistic and Machine Learning Methods

Paper Number: 127340

127340 - Probabilistic Modelling Geometric Tolerance and LCF Life of Gas Turbine Compressor Blade 

Compressor blade geometric tolerances are often typically derived from engineering experience and design precedent. Understanding the corelation between the tolerances at critical locations and cyclic life greatly benefits defining the desirable geometric tolerance that is best for manufacturability, production cost and life.

Many papers on the probabilistic analysis of compressor blade geometry deviations focus on frequency tuning, or some general tolerance correlated with stress using measurements of CMM or 3D scanning. In this paper studies are reported on the contact flange profile tolerance effect on blade LCF life.  The studies address the uncertainties of different geometric deviation patterns and material properties using probabilistic design techniques. This is believed to be the first reported study of this type. This study made great effort through statistical analysis of CMM measurements, material tests and FE analysis to finally give a clear specification for profile tolerance, enabling a relaxation of the current tight tolerance. This greatly benefits the manufacturing process by reducing root manufacturing costs and the additional inspection times arising from over cautious tolerance control.

In this paper typical manufacturing geometric deviations for the root contact flank faces of compressor blades are identified according to the coordinate-measuring machine (CMM) results and patterns are characterised in accordance with the manufacturing process used. The geometric tolerances corresponding to the geometric deviation patterns are illustrated.

FEA models for modelling of these characterised patterns with extreme geometry tolerances are created by modifying only the features corresponding to the blade root contact flank and radius of the compressor blade FE model with normal geometry. A macro based on coordinates transformation are developed to modify the surface node coordinates. FE static elastic-plastic stress analysis with load cycle is performed to achieve the strain range between load cycles. Based on the S-N curves derived by material test and strain range from FE analysis, the LCF life of the compressor blade with various geometric deviations, blade masses, material properties are analysed. The derived LCF lives are used to create the response surface for the probabilistic modelling.  

The probabilistic analysis for the LCF life is carried out by taking the parametrised deviation patterns, blade mass and material properties as random variables. The Monte Carlo simulations are used for probabilistic model. The profile tolerance is established for the root contact flank of compressor blade according to the probabilistic failure criteria.

Presenting Author: Zhiqiang (David) Meng Siemens Energy Industrial Turbomachinery Ltd, Lincoln, United Kingdom

Presenting Author Biography: Zhiqing (David) Meng received his Ph.D.degree (2002) from the Xi'an Jiaotong University, China in mechanical engineering. After 3 years post-doc research at Manchester University and Shanghai Jiaotong University, Dr. Meng has worked in the compressor for the commercial air conditioning and turbomachinery industry for last fifteen years. Since 2012, He has been a principal engineer, lead engineer and then technical specialist at Siemens Energy Industrial Turbomachinery Ltd in Lincoln, UK. He has been working in rotor-dynamics, stress and lifing, probabilistic design for the gas turbine.

Dr. Meng is a CEng and a member of IMechE.

Authors:

Zhiqiang (David) Meng Siemens Energy Industrial Turbomachinery Ltd, Lincoln, United Kingdom
Richard Bluck Siemens Energy Industrial Turbomachinery Ltd, Lincoln, United Kingdom
Björn Sjödin Siemens Energy AB