Robust Rotor Dynamic Design Evaluation of Turbine-Generators Integrating Sensitivity, Uncertainty, and Optimization

Abstract Turbine-generator shaft-trains typically adhere to standard designs based on established modules of turbine and generator rotors. However, contract specifications often necessitate modifications to enhance thermodynamic or rotor dynamic characteristics. This paper presents several case studies where bearing or/and rotor modifications were implemented to meet specific API/ISO standards and contractor requirements. These modifications can be expensive time-consuming and costly, relying heavily on the experience of rotordynamic engineers. Challenges arise, particularly with slender rotor designs, where stability indices, response amplitudes, and other requirements are difficult to achieve. To circumvent the traditional trial-and-error approach, this paper proposes a straightforward alternative method for selecting rotor and bearing modifications. The paper involves a dynamic model of a realistic turbine-generator shaft-train, where sensitivity analysis is first conducted to identify correlations between design inputs (bearing and rotor segment configuration) and design outputs (dynamic characteristics of lateral and torsional vibrations). The procedure also considers integrity and operability aspects such as bearing specific load, power loss, and torsional and bending stress in rotor segments. The objective functions are defined directly from the design outputs, including any API/ISO or contractual restrictions, and are subjected to multi-objective optimization utilizing genetic algorithms. Starting from an existing (rejected) design, the optimization procedure renders the final (acceptable) design within short time with low computational requirements, as linear dynamic models are deployed. The efficiency of the method is demonstrated through various case studies with different restrictions and design targets. Additionally, the method accounts for uncertainty in bearing performance and rotor dynamic calculations, enhancing the reliability of the final design.