Probabilistic Mechanical Design of Turbines
Project director: | Prof. Dr.-Ing. Konrad Vogeler |
Staff: | Dipl.-Ing. Matthias Voigt |
Scientific cooperation: | AG Turbo II GuD 500 MW auf einer Welle |
Foundation: | Rolls-Royce Deutschland Ltd & Co KG , MTU Aero Engines AG, ALSTOM Power GmbH , BMWi |
Duration: | 11/01 - 12/04 |
Probabilistic design methods, why?
Classic turbine design is done using constant, deterministic input parameters for the models of the aerodynamics, cooling system, and structure mechanics. As in every design step conservative assumptions have to be made, the entirety of this assumptions leads to a far too conservative judgement of the components lifetime.
In contrast probabilistic design methods include the scatters of the significant input parameters. Therefore, unlike using classical methods, not only a single, conservative value for the lifetime but a distribution of values based on the probabilistic input parameters can be calculated. Furthermore a sensitivity analysis can be used to identify parameters critical for the lifetime. These parameters can be optimised afterwards.
What has to be researched?
The goal of this project is to include the probabilistic character of certain parameters in turbine design in the process of development and optimisation. Work has to be done for the following subtasks:
Assortment of adequate probabilistic methods and software tools
Which probabilistic methods are available for sensitivity analysis (Monte- Carlo- Simulation, Response Surface Method, sensitivity analysis using the rank order correlation)?
- Which software tools are available (e.g. ST-ORM, iSIGHT)? What are there strengths and limitations?
- Assortment of adequate stochastic variables and models to characterize them
- Variables connected to cooling (e.g. hydraulic models of the blade cooling, heat transfer coefficients and temperatures)
- Manufacturing tolerances (e.g. variation of wall thickness, geometric variations of the cooling system)
- Variables in the models for the material (e.g. Youngs modulus, coefficient of linear thermal expansion)
- Improvement of the efficiency of probabilistic design methods
- Methods to reduce the number of stochastic variables
- Methods to reduce the number of deterministic calculations
- Parallel computing / super computing
The recognition and modelling of interdisciplinary coherences is essential for the application of probabilistic methods. Therefore, besides the challenges mentioned above, an integrated interdisciplinary approach for the design process is necessary. This in term is an important step towards an interdisciplinary overall optimisation of the Total Life Cycle Cost of the turbine components analysed.
Publications within this project
Matthias Voigt, Roland Mücke, M. Oevermann, K. Vogeler
Probablistic Lifetime Analysis for Turbine Blades Based on a Combined Direct Monte-Carlo and Respond Surface Approach
Draft ASME Turbo Expo GT2004-53439, 49th ASME International Gas Turbine & Aeroengine Technical Congress & Exposition, Vienna, Austria, June 14-17, 2004
Other publications are available on the german page.