0
Technical Briefs

Application of Artificial Neural Networks in Investigations of Steam Turbine Cascades

[+] Author and Article Information
Krzysztof Kosowski, Karol Tucki

Department of Ship Automatics and Turbine Propulsion, Gdańsk University of Technology, Narutowicza 11/12, 80952 Gdańsk, Poland

Adrian Kosowski

Department of Algorithms and System Modeling, Gdańsk University of Technology, Narutowicza 11/12, 80952 Gdańsk, Poland

Tests were performed for nozzle profiles of PLK and PLK-V types, and rotor blade profiles of P2 and P2-V types, according to the ALSTOM Power notation.

J. Turbomach 132(1), 014501 (Sep 11, 2009) (5 pages) doi:10.1115/1.3103923 History: Received February 17, 2008; Revised January 24, 2009; Published September 11, 2009

We present the results of numerical tests of artificial neural networks (ANNs) applied in the investigations of flows in steam turbine cascades. Typical constant cross-sectional blades, as well as high-performance blades, were both considered. The obtained results indicate that ANNs may be used for estimating the spatial distribution of flow parameters, such as enthalpy, entropy, pressure, velocity, and energy losses, in the flow channel. Finally, we remark on the application of ANNs in the design process of turbine flow parts, as an extremely fast complementary method for many 3D computational fluid dynamics calculations. By using ANNs combined with evolutionary algorithms, it is possible to reduce by several orders of magnitude the time of design optimization for cascades, stages, and groups of stages.

FIGURES IN THIS ARTICLE
<>
Copyright © 2010 by American Society of Mechanical Engineers
Your Session has timed out. Please sign back in to continue.

References

Figures

Grahic Jump Location
Figure 1

Geometry of a turbine cascade

Grahic Jump Location
Figure 2

Schematic of a feedforward network (according to Ref. 15)

Grahic Jump Location
Figure 3

A comparison of enthalpy distribution predicted using ANN and calculated by the FlowER solver at the axial section at 62% of the rotor blade channel

Grahic Jump Location
Figure 4

A comparison of enthalpy distribution predicted using ANN and calculated by the FlowER solver at the section at 30% of nozzle channel height, counting from the hub

Grahic Jump Location
Figure 5

A comparison of velocity component distribution predicted using ANN and calculated by the FlowER solver at the section at 29% of nozzle channel height, counting from the hub

Grahic Jump Location
Figure 6

Variants of nozzle modification: (a) straight lean, (b) compound lean, and (c) compound sweep

Grahic Jump Location
Figure 7

Examples of relations between energy losses ζ and volume flow rate V at the stage outlet, used for training (solid lines; computed by CFD code) and testing (dotted lines) the ANN for different types of cascades: (a) rotor blades with straight lean, (b) nozzles with compound lean, and (c) nozzles with compound sweep. Test results obtained from ANN and CFD code for the same parameters are indistinguishable in the graph.

Grahic Jump Location
Figure 8

Relative prediction errors of proposed neural networks for a test set of 32 real world cascade arrangements (solid line—prediction of cascade losses ζ; dashed line—prediction of blade height l)

Grahic Jump Location
Figure 9

Example of overall efficiency η of particular stages and schema of rotor blades of a group of six steam turbine stages designed using ANN and evolutionary algorithms

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In